1. daniel archambaultmaterials.dagstuhl.de/files/15/15481/15481.swm1.other.pdf · platform...

116
Running order Daniel Archambault Benjamin Bach Kathrin Ballweg Rita Borgo Alessandro Bozzon Sheelagh Carpendale Remco Chang Min Chen Stephan Diehl Darren Edwards Sebastian Egger Sara Fabrikant Brian Fisher Ujwal Gadiraju Neha Gupta Matthias Hirth Tobias Hoßfeld Jason Jacques Radu Jianu Christian Keimel Andreas Kerren Stephen Kobourov Bongshin Lee David Martin Andrea Mauri Fintan McGee Luana Micallef Sebastian Möller Babak Naderi Martin Nöllenburg Helen Purchase Judith Redi Peter Rodgers Dietmar Saupe Ognjen Šćekić Paolo Simonetto Tatiana von Landesberger Ina Wechsung Michael Wybrow Michelle Zhou

Upload: others

Post on 30-Apr-2020

2 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Running order Daniel Archambault

Benjamin Bach

Kathrin Ballweg

Rita Borgo

Alessandro Bozzon

Sheelagh Carpendale

Remco Chang

Min Chen

Stephan Diehl

Darren Edwards

Sebastian Egger

Sara Fabrikant

Brian Fisher

Ujwal Gadiraju

Neha Gupta

Matthias Hirth

Tobias Hoszligfeld

Jason Jacques

Radu Jianu

Christian Keimel

Andreas Kerren

Stephen Kobourov

Bongshin Lee

David Martin

Andrea Mauri

Fintan McGee

Luana Micallef

Sebastian Moumlller

Babak Naderi

Martin Noumlllenburg

Helen Purchase

Judith Redi

Peter Rodgers

Dietmar Saupe

Ognjen Šćekić

Paolo Simonetto

Tatiana von Landesberger

Ina Wechsung

Michael Wybrow

Michelle Zhou

Daniel Archambault

Daniel Archambault

bull Lecturer Swansea University

bull Research Interests

bull Information Visualisation

bull Graph Drawing and Visualisation

bull Perceptual Issues

bull Scalability Issues

bull Human-Centred Methodology to Evaluate Visualisations

Seminar Interests

bull Human-Centred Evaluation of Visualisations

bull How to Effectively and Correctly Use Crowdsourcing

bull Experimental methodology (between vs within)

bull When not to use crowdsourcing as well as when to use it

Benjamin Bach

Benjamin Bach Microsoft Research ndash Inria Joint Centre Saclay France

Temporal Data Dynamic Networks Domain Collaborations

Benjamin Bach Microsoft Research ndash Inria Joint Centre Saclay France

How to motivate experts in participating

What kind of feedback can we expect

Which infrastructure do we need

How to find interested domain experts

How to design (micro) tasks

Kathrin Ballweg

Kathrin Ballweg ndash About me

Interests

Computer Science

Cognitive Psychology

User Studies

PhD Focus

Perception and cognition in

network visualization

Visualization for the masses

(journalists and public ndash news readers)

231115 | Dagstuhl Seminar ndash Evaluation in the Crowd Crowdsouring and Human-Centered Experiments | K Ballweg | 9

Kathrin Ballweg

M Sc Computer Science

Technische Universitaumlt

Darmstadt

Design

Evaluation in the Crowd Crowdsourcing and

Human-Centered Experiments

Topics of Interest

Crowdsourcing Platforms vs

The Laboratory

Scientifically Rigorous Methodologies

Crowdsourcing Experiments in Human-

Computer Interaction Visualization and

Applied PerceptionGraphics

231115 | Dagstuhl Seminar ndash Evaluation in the Crowd Crowdsouring and Human-Centered Experiments | K Ballweg | 10

httpwwwtagesnetzwerkde

Rita Borgo

Introduction

Rita Borgo

Visual Computing

Swansea University

Large Data Visualisation HPC

Glyph Based Visualisation

Time Series Analysis

Perception in Visualisation

Visual Computing Group Swansea University

Prof MWJones

D Archamabault R Borgo

R S Laramee B Mora

K W Tam X Xie

My Interest in the Crowdsourcing Phenomena

bull Crowdsourcing is a precious and very appealing resource

bull Crowdsourcing is a powerful instrument

bull However it is an instrument we do not know thoroughly yet we use ithellip

Therefore

1 Can we help each other to identify ldquoguidelinesrdquo to help us use this instrument in the best way an to its full potentials

2 Is there anything WE can DO to make it even better

Alessandro Bozzon

Alessandro Bozzon

Assistant Professor TU Delft Human-Enhanced Data Management Web Information Systems Social Data Science

Faculty Fellow IBM Benelux Inclusive Enterprise

Investigator AMS Social Sensing Smart Citizens

Web Engineering Web Science

Information Retrieval

User Modelling Crowdsourcing

Human Computation

WWW

ICWE

UMAP

HT

VLDBJ

TWEB

IC

SWJ

JWE

JWS

CSCW

ISWC

Publication Venues

wwwalessandrobozzoncom

Topics of Interest

How can humans and machines better collaborate in solving

(computational) problems

Machines Social

Data

People

Process

A s

ocio

-tech

nic

al syste

m How can human-generated

Web data be transformed

into a source that informs

Web system design

How to enhance Web-

based systems with

automated large-scale

human interpretation

GO

AL

S Correctness

Task Design

Experiment Methodologies

Efficiency Worker Modeling

Getting to Know the Crowd

Sustainability Incentives

Getting to Know the Crowd

Sheelagh Carpendale

Sheelagh Carpendale Canada Research Chair Information Visualization

NSERCAITFSMART Industrial Research Chair Interactive Technologies

InnoVis (Innovations in Visualization)

Interactions Lab

Computer Science University of Calgary

Interactive

Information

Visualization

I have come here to learn

bull While I do a lot of empirical research most is qualitative

bull While I think that crowd sourcing is potentially a fantastic new resource I think successful use of crowd sourcing is non-trivial

bull I hope to learn about how to work with this

Remco Chang

21 Remco Chang ndash Dagstuhl 15

From vision science to data science applying perception to problems in big data

Remco Chang

Assistant Professor Computer Science

Tufts University

22 Remco Chang ndash Dagstuhl 15

Weberrsquos Law

119889119875 = 119896119889119878

119878

Min Chen

Min Chen Professor of Scientific Visualization

University of Oxford (2011 ~ Present) Swansea University (1984 ~ 2011)

Research Interests most aspects of visualization aspects of HCI computer graphics computer vision data mining and information theory

Research Highlight

1 Volume Graphics and Visualization

2 Video Visualization

3 Visual Analytics

4 Perception and Cognition in Visualization

5 Theory of Visualization

1 2

3 4

5

Research Question in the Context of 15481

Short Version How can we use crowdsourcing to capture perceptual and cognitive measures to facilitate mathematical modelling of visualization and interaction

Long Version Information theory is perhaps the most promising mathematical framework that can underpin the fields of visualization and interaction as well as data processing and communication However to obtain its core measures various types of entropy one needs to capture a probability distribution of all states associated with a concept While such a probability distribution can be captured in a machine-centric data processing or communication pipeline it is not trivial to capture in a human-centric process Crowdsourcing offers an opportunity to conduct ldquobigrdquo observation and measurement of perceptual and cognitive activities It is thus a tool for ldquoBig Sciencerdquo

Stephan Diehl

Dagstuhl Seminar 15481 Evaluation in the Crowd Crowdsourcing and Human-Centred Experiments

Stephan Diehl Software Engineering University of Trier Germany

The Java Code Clone

Detection API

Empirical Software Engineering bull [JSS2015 ESEM2015 FSE2014 ICSM 2012

FSE 2011 ESEM 2011 hellip]

Empirical Information Visualization bull [InfoVis 2010 J of Media Psychology 410 hellip]

Analysis and Collaboration SE Tools bull [ASE2014 ASE 2012 ICSE 2011 TSE 2005 ]

Visualizing (Dynamic) Compound Graphs bull [EuroVisSTAR 2014 IV Journal 2012 InfoVis11

SOFTVIS10 EuroVis09 IV09 EuroVis08

CHASE08 AVI08GD04GD02 hellip]

Dagstuhl Seminar 15481 Evaluation in the Crowd Crowdsourcing and Human-Centred Experiments

bull How to reach your (very specific) target audience without

spamming

bull How to control for resolution animation speed color

bull What crowdsourcing platforms exist and what are their

features

Darren Edwards

Darren Edwards Swansea University

Cognitive psychology - categorization

Health and clinical psychology ndash applications of categorization on clinical populations (eg autism)

Crowdsourcing and psychological interventions

Recent validity studies confirm the use of simple psychological tasks such as decision making

Recent attempts have now explored more complex tasks such as psychological interventions (eg mindfulness)

Present studies now explore more complex forms of mindfulness

What are the limitations of using crowdsourced populations online such as when using psychological interventions or categorization

Sebastian Egger

Engineering

Technical high school

bull Telematics

electrotechnical enginieering telecommunications

bull Signal Processing for UWB systems

Sociology

Theory of human societies

Interaction amp collaboration behaviour of groups

Measurement of human behaviour

Currently Human Computer Interaction

Social amp collaborative aspects of technology use

33 26112015

Sebastian Egger

Motivational aspects of crowdworkers

CS as a collaborative work setting

Optimal Setup of CS experiments

Questions (max Nr of questions max duration)

Gamified reliability checks

Scale designs

CS vs Lab whatlsquos the ground truth

New application domains of CS

34 26112015

Interest in Crowdsourcing

Sara Fabrikant

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

humanmdashsystem (visuo-spatial) interface

spatio-temporal analytics

spatialization human navigation

multivariate spatial analysis etc)

interface design

large amp small interactive (map) displays

(ie desktop mobile VR etc)

empirical evaluations

(ie eye tracking EDA EEG etc) [httpwwwunitsevenconz]

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

human sensing observatory set up

controlled laboratory | in-situ (messy) real world setting

semi|automatic data collection of various human behavioral data streams

ie psycho-physiological perceptual cognitive and sensory-motor behaviors (through crowdsourcing )

Brian Fisher

My double research life

Brian Fisher

bull PhD in Experimental Psych

bull Postdoc w Cognitive Science

society founder amp president

bull Psychonomics Fellow

bull VIS-related symposia at Cogsci

amp APS papers on cogsci of

interaction

bull Fuzzy-logicBayes models

bull Postdoc funded by Inst for Robotics

and AI

bull VAST SC VEC VACCINE

bull Co-organized Dagstuhl ldquoInteraction

with Information for Visual Reasoningrdquo

Cogsci-based papers at VIS CHI

BELIV

My Plan

VIS

Me Psychological Science

Ujwal Gadiraju

p = 0055111

Crowd Worker

Behavior

Microtask

Design

Crowdsourcing Paradigm Quality amp Effectiveness

UJWAL GADIRAJU

Incentivization Gamification

Neha Gupta

Getting to know the crowd

Presented at Dagstuhl by

Dr David Martin amp Neha Gupta

Neha Gupta

PhD Student School of Computer Science University of Nottingham UK

Investigating crowdwork through the platform Amazon Mechanical Turk

psxng1nottinghamacuk

Neha_Gi_ji

Research amp Research teams

Research Studies

bull Qualitative study of Turkers on online forums like mturkforum and Turker Nation by Dr David Martin XRCE

bull Ethnographic Studies of Turkers in their respective homes and places of work in India by Neha Gupta UoN

Research teams

Xerox

bull David Martin Jacki OrsquoNeill (Now MSR India) Ben Hanrahan (Now PSU)

psxng1nottinghamacuk

Neha_Gi_ji

Nottingham

bull Andy Crabtree

Tom Rodden

Matthias Hirth

Matthias Hirth 48 48

About Myself

Matthias Hirth

Research Assistant at University of Wuumlrzburg

httpmatthias-hirthcom

matthiashirthinformatikuni-wuerzburgde

Crowdsourcing

Platform analysis

Platform optimization

Human factors

Use cases

Online Social Networks

Collaboration networks

Structural differences of OSNs

Application of OSN relationships

Interactive applicationsGaming

Challenges of real time interactions

Serious gamingGamification

E-Sports broadcastinglive streaming

Quality of Experience

Monitoring techniques

Quantification techniques

(Crowd-bases) test design

Social interactions

Test methodology Incentive design

Use case

Measurement technique

Identification of

qualification

Matthias Hirth 49 49

Improving Crowdsourcing Technology and the Way We Use it

Crowdsourcing technology

How to transfer well known lab procedures into an uncontrolled

crowdsourcing environment

Which crowdsourcing platform features are required to conduct

successful experiments but are not present yet

Which technical developments (can) foster the usage of

crowdsourcing instead of lab experiments

How can technological developments be used for novel incentive

mechanisms

Usage of Crowdsourcing

How can we conduct reproducible experiments in diverse

crowdsourcing environment

How can we raise the awareness of crowdsourcing employers for the

individuals behind the crowd

Tobias Hoszligfeld

The crowd community in the lab

Tobias Hoszligfeld wwwmaswiwiuni-duede

bull Background Tobias Hoszligfeld

ndash 2004-2014 Research group ldquoFuture Internet Applicationsrdquo Chair of Communication Networks Univ of Wuumlrzburg

ndash Since Sep 2014 Chair of Modeling of Adaptive Systems (head) University of Duisburg-Essen

bull Interests Application of crowdsourcing

ndash User studies on Quality of Experience (QoE) for Internet apps eg video QoE in the crowd amp in the lab methodology reliability statisticshellip [2]

ndash Mobile crowdsourcing network and context measurements eg Wifi offloading incentives privacy recommendation mechanisms etc [1]

bull Interests The crowd community

ndash Crowd = Human beings Social network wisdom of the crowd community

ndash UnderstandingUtilizing the crowd community for successful human computation For which jobs can social interaction improve the quality of work Classification

ndash How can we deploy such a system where interaction is triggered

ndash System design Task design interaction mechanisms incentives

22 Nov 2015

[1] httpdropsdagstuhldeopusvolltexte20134354

[2] httpswww3informatikuni-wuerzburgdeqoewikiqualinetcrowd

Challenges Example of Crowdsourced QoE

Tobias Hoszligfeld wwwmaswiwiuni-duede

task design

incentives

Internet

crowd

community

anonymous

remote

subjects

Internet delivery of

test data and user

response

QoE study bull testing methodology bull QoE influence factors

under investigation bull additional QoE factors bull proper subset of items bull hellip

crowdsourcing setting bull short task duration bull instructions amp training bull feedback amp questions bull hidden influence factors bull reliability of users bull hellip

heterogeneous devices

and Internet access

implementation extrinsic vs intrinsic

motivation

Processing of data

Technical aspects bull server implementation bull client application amp

compatibility bull desired test conditions bull context monitoring bull hellip

test items

lab

scientific methodologies

22 Nov 2015

Jason Jacques

Jason Jacques supervised by Per Ola Kristensson

T

Crowdsourcing Code Perception

jtj21camacuk peopledscamacukjtj21

jtjacquesgmailcom jsonjcouk

PhD Working Title

science

Crowdsourcing Interests W

hat

W

ho

Ho

w

Wh

y jtj21camacuk jsonjcouk science

Radu Jianu

Radu Jianu Assistant Professor

Florida International University (PhD Brown University lsquo12)

Interests

Applications of eye-tracking in visualization tracking what people look at rather than where they look

bull Data intensive large scale eye-tracking experiments bull Gaze-contingent visualizations bull Gaze-driven data recommendation and big data curation

Automating user studies in data visualizations (on next page)

Automating user studies

1 Software designs that can (semi)automate a broad range of quantitative evaluations by leveraging

ndash Crowdsourcing ndash Benchmark data and task taxonomies ndash Automated result interpretation

2 Moving user evaluation from the end to within the software design and implementation process as a means to choose between competitive visual designs

Christian Keimel

copy IRT

Where

What

bull Hybrid broadcast broadband technologies

bull Big Data in broadcasting environment and content-creation

bull Audio-visual QoE assessment (with crowdsourcing)

About me

Christian Keimel Slide 60

Lecturer for Digital Broadcast Engineering

Research Engineer for Media Services and Platforms

copy IRT

Crowdsourcing in large scale optimisation

The human component of Big Data

Moving crowdsourcing to lean-back devices

Crowdsourcing from the couch

Crowdsourced experiments emulating the laboratory

Is it time for new methodologies ignoring the lab

Questionshellip

Christian Keimel Slide 61

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 2: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Daniel Archambault

Daniel Archambault

bull Lecturer Swansea University

bull Research Interests

bull Information Visualisation

bull Graph Drawing and Visualisation

bull Perceptual Issues

bull Scalability Issues

bull Human-Centred Methodology to Evaluate Visualisations

Seminar Interests

bull Human-Centred Evaluation of Visualisations

bull How to Effectively and Correctly Use Crowdsourcing

bull Experimental methodology (between vs within)

bull When not to use crowdsourcing as well as when to use it

Benjamin Bach

Benjamin Bach Microsoft Research ndash Inria Joint Centre Saclay France

Temporal Data Dynamic Networks Domain Collaborations

Benjamin Bach Microsoft Research ndash Inria Joint Centre Saclay France

How to motivate experts in participating

What kind of feedback can we expect

Which infrastructure do we need

How to find interested domain experts

How to design (micro) tasks

Kathrin Ballweg

Kathrin Ballweg ndash About me

Interests

Computer Science

Cognitive Psychology

User Studies

PhD Focus

Perception and cognition in

network visualization

Visualization for the masses

(journalists and public ndash news readers)

231115 | Dagstuhl Seminar ndash Evaluation in the Crowd Crowdsouring and Human-Centered Experiments | K Ballweg | 9

Kathrin Ballweg

M Sc Computer Science

Technische Universitaumlt

Darmstadt

Design

Evaluation in the Crowd Crowdsourcing and

Human-Centered Experiments

Topics of Interest

Crowdsourcing Platforms vs

The Laboratory

Scientifically Rigorous Methodologies

Crowdsourcing Experiments in Human-

Computer Interaction Visualization and

Applied PerceptionGraphics

231115 | Dagstuhl Seminar ndash Evaluation in the Crowd Crowdsouring and Human-Centered Experiments | K Ballweg | 10

httpwwwtagesnetzwerkde

Rita Borgo

Introduction

Rita Borgo

Visual Computing

Swansea University

Large Data Visualisation HPC

Glyph Based Visualisation

Time Series Analysis

Perception in Visualisation

Visual Computing Group Swansea University

Prof MWJones

D Archamabault R Borgo

R S Laramee B Mora

K W Tam X Xie

My Interest in the Crowdsourcing Phenomena

bull Crowdsourcing is a precious and very appealing resource

bull Crowdsourcing is a powerful instrument

bull However it is an instrument we do not know thoroughly yet we use ithellip

Therefore

1 Can we help each other to identify ldquoguidelinesrdquo to help us use this instrument in the best way an to its full potentials

2 Is there anything WE can DO to make it even better

Alessandro Bozzon

Alessandro Bozzon

Assistant Professor TU Delft Human-Enhanced Data Management Web Information Systems Social Data Science

Faculty Fellow IBM Benelux Inclusive Enterprise

Investigator AMS Social Sensing Smart Citizens

Web Engineering Web Science

Information Retrieval

User Modelling Crowdsourcing

Human Computation

WWW

ICWE

UMAP

HT

VLDBJ

TWEB

IC

SWJ

JWE

JWS

CSCW

ISWC

Publication Venues

wwwalessandrobozzoncom

Topics of Interest

How can humans and machines better collaborate in solving

(computational) problems

Machines Social

Data

People

Process

A s

ocio

-tech

nic

al syste

m How can human-generated

Web data be transformed

into a source that informs

Web system design

How to enhance Web-

based systems with

automated large-scale

human interpretation

GO

AL

S Correctness

Task Design

Experiment Methodologies

Efficiency Worker Modeling

Getting to Know the Crowd

Sustainability Incentives

Getting to Know the Crowd

Sheelagh Carpendale

Sheelagh Carpendale Canada Research Chair Information Visualization

NSERCAITFSMART Industrial Research Chair Interactive Technologies

InnoVis (Innovations in Visualization)

Interactions Lab

Computer Science University of Calgary

Interactive

Information

Visualization

I have come here to learn

bull While I do a lot of empirical research most is qualitative

bull While I think that crowd sourcing is potentially a fantastic new resource I think successful use of crowd sourcing is non-trivial

bull I hope to learn about how to work with this

Remco Chang

21 Remco Chang ndash Dagstuhl 15

From vision science to data science applying perception to problems in big data

Remco Chang

Assistant Professor Computer Science

Tufts University

22 Remco Chang ndash Dagstuhl 15

Weberrsquos Law

119889119875 = 119896119889119878

119878

Min Chen

Min Chen Professor of Scientific Visualization

University of Oxford (2011 ~ Present) Swansea University (1984 ~ 2011)

Research Interests most aspects of visualization aspects of HCI computer graphics computer vision data mining and information theory

Research Highlight

1 Volume Graphics and Visualization

2 Video Visualization

3 Visual Analytics

4 Perception and Cognition in Visualization

5 Theory of Visualization

1 2

3 4

5

Research Question in the Context of 15481

Short Version How can we use crowdsourcing to capture perceptual and cognitive measures to facilitate mathematical modelling of visualization and interaction

Long Version Information theory is perhaps the most promising mathematical framework that can underpin the fields of visualization and interaction as well as data processing and communication However to obtain its core measures various types of entropy one needs to capture a probability distribution of all states associated with a concept While such a probability distribution can be captured in a machine-centric data processing or communication pipeline it is not trivial to capture in a human-centric process Crowdsourcing offers an opportunity to conduct ldquobigrdquo observation and measurement of perceptual and cognitive activities It is thus a tool for ldquoBig Sciencerdquo

Stephan Diehl

Dagstuhl Seminar 15481 Evaluation in the Crowd Crowdsourcing and Human-Centred Experiments

Stephan Diehl Software Engineering University of Trier Germany

The Java Code Clone

Detection API

Empirical Software Engineering bull [JSS2015 ESEM2015 FSE2014 ICSM 2012

FSE 2011 ESEM 2011 hellip]

Empirical Information Visualization bull [InfoVis 2010 J of Media Psychology 410 hellip]

Analysis and Collaboration SE Tools bull [ASE2014 ASE 2012 ICSE 2011 TSE 2005 ]

Visualizing (Dynamic) Compound Graphs bull [EuroVisSTAR 2014 IV Journal 2012 InfoVis11

SOFTVIS10 EuroVis09 IV09 EuroVis08

CHASE08 AVI08GD04GD02 hellip]

Dagstuhl Seminar 15481 Evaluation in the Crowd Crowdsourcing and Human-Centred Experiments

bull How to reach your (very specific) target audience without

spamming

bull How to control for resolution animation speed color

bull What crowdsourcing platforms exist and what are their

features

Darren Edwards

Darren Edwards Swansea University

Cognitive psychology - categorization

Health and clinical psychology ndash applications of categorization on clinical populations (eg autism)

Crowdsourcing and psychological interventions

Recent validity studies confirm the use of simple psychological tasks such as decision making

Recent attempts have now explored more complex tasks such as psychological interventions (eg mindfulness)

Present studies now explore more complex forms of mindfulness

What are the limitations of using crowdsourced populations online such as when using psychological interventions or categorization

Sebastian Egger

Engineering

Technical high school

bull Telematics

electrotechnical enginieering telecommunications

bull Signal Processing for UWB systems

Sociology

Theory of human societies

Interaction amp collaboration behaviour of groups

Measurement of human behaviour

Currently Human Computer Interaction

Social amp collaborative aspects of technology use

33 26112015

Sebastian Egger

Motivational aspects of crowdworkers

CS as a collaborative work setting

Optimal Setup of CS experiments

Questions (max Nr of questions max duration)

Gamified reliability checks

Scale designs

CS vs Lab whatlsquos the ground truth

New application domains of CS

34 26112015

Interest in Crowdsourcing

Sara Fabrikant

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

humanmdashsystem (visuo-spatial) interface

spatio-temporal analytics

spatialization human navigation

multivariate spatial analysis etc)

interface design

large amp small interactive (map) displays

(ie desktop mobile VR etc)

empirical evaluations

(ie eye tracking EDA EEG etc) [httpwwwunitsevenconz]

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

human sensing observatory set up

controlled laboratory | in-situ (messy) real world setting

semi|automatic data collection of various human behavioral data streams

ie psycho-physiological perceptual cognitive and sensory-motor behaviors (through crowdsourcing )

Brian Fisher

My double research life

Brian Fisher

bull PhD in Experimental Psych

bull Postdoc w Cognitive Science

society founder amp president

bull Psychonomics Fellow

bull VIS-related symposia at Cogsci

amp APS papers on cogsci of

interaction

bull Fuzzy-logicBayes models

bull Postdoc funded by Inst for Robotics

and AI

bull VAST SC VEC VACCINE

bull Co-organized Dagstuhl ldquoInteraction

with Information for Visual Reasoningrdquo

Cogsci-based papers at VIS CHI

BELIV

My Plan

VIS

Me Psychological Science

Ujwal Gadiraju

p = 0055111

Crowd Worker

Behavior

Microtask

Design

Crowdsourcing Paradigm Quality amp Effectiveness

UJWAL GADIRAJU

Incentivization Gamification

Neha Gupta

Getting to know the crowd

Presented at Dagstuhl by

Dr David Martin amp Neha Gupta

Neha Gupta

PhD Student School of Computer Science University of Nottingham UK

Investigating crowdwork through the platform Amazon Mechanical Turk

psxng1nottinghamacuk

Neha_Gi_ji

Research amp Research teams

Research Studies

bull Qualitative study of Turkers on online forums like mturkforum and Turker Nation by Dr David Martin XRCE

bull Ethnographic Studies of Turkers in their respective homes and places of work in India by Neha Gupta UoN

Research teams

Xerox

bull David Martin Jacki OrsquoNeill (Now MSR India) Ben Hanrahan (Now PSU)

psxng1nottinghamacuk

Neha_Gi_ji

Nottingham

bull Andy Crabtree

Tom Rodden

Matthias Hirth

Matthias Hirth 48 48

About Myself

Matthias Hirth

Research Assistant at University of Wuumlrzburg

httpmatthias-hirthcom

matthiashirthinformatikuni-wuerzburgde

Crowdsourcing

Platform analysis

Platform optimization

Human factors

Use cases

Online Social Networks

Collaboration networks

Structural differences of OSNs

Application of OSN relationships

Interactive applicationsGaming

Challenges of real time interactions

Serious gamingGamification

E-Sports broadcastinglive streaming

Quality of Experience

Monitoring techniques

Quantification techniques

(Crowd-bases) test design

Social interactions

Test methodology Incentive design

Use case

Measurement technique

Identification of

qualification

Matthias Hirth 49 49

Improving Crowdsourcing Technology and the Way We Use it

Crowdsourcing technology

How to transfer well known lab procedures into an uncontrolled

crowdsourcing environment

Which crowdsourcing platform features are required to conduct

successful experiments but are not present yet

Which technical developments (can) foster the usage of

crowdsourcing instead of lab experiments

How can technological developments be used for novel incentive

mechanisms

Usage of Crowdsourcing

How can we conduct reproducible experiments in diverse

crowdsourcing environment

How can we raise the awareness of crowdsourcing employers for the

individuals behind the crowd

Tobias Hoszligfeld

The crowd community in the lab

Tobias Hoszligfeld wwwmaswiwiuni-duede

bull Background Tobias Hoszligfeld

ndash 2004-2014 Research group ldquoFuture Internet Applicationsrdquo Chair of Communication Networks Univ of Wuumlrzburg

ndash Since Sep 2014 Chair of Modeling of Adaptive Systems (head) University of Duisburg-Essen

bull Interests Application of crowdsourcing

ndash User studies on Quality of Experience (QoE) for Internet apps eg video QoE in the crowd amp in the lab methodology reliability statisticshellip [2]

ndash Mobile crowdsourcing network and context measurements eg Wifi offloading incentives privacy recommendation mechanisms etc [1]

bull Interests The crowd community

ndash Crowd = Human beings Social network wisdom of the crowd community

ndash UnderstandingUtilizing the crowd community for successful human computation For which jobs can social interaction improve the quality of work Classification

ndash How can we deploy such a system where interaction is triggered

ndash System design Task design interaction mechanisms incentives

22 Nov 2015

[1] httpdropsdagstuhldeopusvolltexte20134354

[2] httpswww3informatikuni-wuerzburgdeqoewikiqualinetcrowd

Challenges Example of Crowdsourced QoE

Tobias Hoszligfeld wwwmaswiwiuni-duede

task design

incentives

Internet

crowd

community

anonymous

remote

subjects

Internet delivery of

test data and user

response

QoE study bull testing methodology bull QoE influence factors

under investigation bull additional QoE factors bull proper subset of items bull hellip

crowdsourcing setting bull short task duration bull instructions amp training bull feedback amp questions bull hidden influence factors bull reliability of users bull hellip

heterogeneous devices

and Internet access

implementation extrinsic vs intrinsic

motivation

Processing of data

Technical aspects bull server implementation bull client application amp

compatibility bull desired test conditions bull context monitoring bull hellip

test items

lab

scientific methodologies

22 Nov 2015

Jason Jacques

Jason Jacques supervised by Per Ola Kristensson

T

Crowdsourcing Code Perception

jtj21camacuk peopledscamacukjtj21

jtjacquesgmailcom jsonjcouk

PhD Working Title

science

Crowdsourcing Interests W

hat

W

ho

Ho

w

Wh

y jtj21camacuk jsonjcouk science

Radu Jianu

Radu Jianu Assistant Professor

Florida International University (PhD Brown University lsquo12)

Interests

Applications of eye-tracking in visualization tracking what people look at rather than where they look

bull Data intensive large scale eye-tracking experiments bull Gaze-contingent visualizations bull Gaze-driven data recommendation and big data curation

Automating user studies in data visualizations (on next page)

Automating user studies

1 Software designs that can (semi)automate a broad range of quantitative evaluations by leveraging

ndash Crowdsourcing ndash Benchmark data and task taxonomies ndash Automated result interpretation

2 Moving user evaluation from the end to within the software design and implementation process as a means to choose between competitive visual designs

Christian Keimel

copy IRT

Where

What

bull Hybrid broadcast broadband technologies

bull Big Data in broadcasting environment and content-creation

bull Audio-visual QoE assessment (with crowdsourcing)

About me

Christian Keimel Slide 60

Lecturer for Digital Broadcast Engineering

Research Engineer for Media Services and Platforms

copy IRT

Crowdsourcing in large scale optimisation

The human component of Big Data

Moving crowdsourcing to lean-back devices

Crowdsourcing from the couch

Crowdsourced experiments emulating the laboratory

Is it time for new methodologies ignoring the lab

Questionshellip

Christian Keimel Slide 61

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 3: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Daniel Archambault

bull Lecturer Swansea University

bull Research Interests

bull Information Visualisation

bull Graph Drawing and Visualisation

bull Perceptual Issues

bull Scalability Issues

bull Human-Centred Methodology to Evaluate Visualisations

Seminar Interests

bull Human-Centred Evaluation of Visualisations

bull How to Effectively and Correctly Use Crowdsourcing

bull Experimental methodology (between vs within)

bull When not to use crowdsourcing as well as when to use it

Benjamin Bach

Benjamin Bach Microsoft Research ndash Inria Joint Centre Saclay France

Temporal Data Dynamic Networks Domain Collaborations

Benjamin Bach Microsoft Research ndash Inria Joint Centre Saclay France

How to motivate experts in participating

What kind of feedback can we expect

Which infrastructure do we need

How to find interested domain experts

How to design (micro) tasks

Kathrin Ballweg

Kathrin Ballweg ndash About me

Interests

Computer Science

Cognitive Psychology

User Studies

PhD Focus

Perception and cognition in

network visualization

Visualization for the masses

(journalists and public ndash news readers)

231115 | Dagstuhl Seminar ndash Evaluation in the Crowd Crowdsouring and Human-Centered Experiments | K Ballweg | 9

Kathrin Ballweg

M Sc Computer Science

Technische Universitaumlt

Darmstadt

Design

Evaluation in the Crowd Crowdsourcing and

Human-Centered Experiments

Topics of Interest

Crowdsourcing Platforms vs

The Laboratory

Scientifically Rigorous Methodologies

Crowdsourcing Experiments in Human-

Computer Interaction Visualization and

Applied PerceptionGraphics

231115 | Dagstuhl Seminar ndash Evaluation in the Crowd Crowdsouring and Human-Centered Experiments | K Ballweg | 10

httpwwwtagesnetzwerkde

Rita Borgo

Introduction

Rita Borgo

Visual Computing

Swansea University

Large Data Visualisation HPC

Glyph Based Visualisation

Time Series Analysis

Perception in Visualisation

Visual Computing Group Swansea University

Prof MWJones

D Archamabault R Borgo

R S Laramee B Mora

K W Tam X Xie

My Interest in the Crowdsourcing Phenomena

bull Crowdsourcing is a precious and very appealing resource

bull Crowdsourcing is a powerful instrument

bull However it is an instrument we do not know thoroughly yet we use ithellip

Therefore

1 Can we help each other to identify ldquoguidelinesrdquo to help us use this instrument in the best way an to its full potentials

2 Is there anything WE can DO to make it even better

Alessandro Bozzon

Alessandro Bozzon

Assistant Professor TU Delft Human-Enhanced Data Management Web Information Systems Social Data Science

Faculty Fellow IBM Benelux Inclusive Enterprise

Investigator AMS Social Sensing Smart Citizens

Web Engineering Web Science

Information Retrieval

User Modelling Crowdsourcing

Human Computation

WWW

ICWE

UMAP

HT

VLDBJ

TWEB

IC

SWJ

JWE

JWS

CSCW

ISWC

Publication Venues

wwwalessandrobozzoncom

Topics of Interest

How can humans and machines better collaborate in solving

(computational) problems

Machines Social

Data

People

Process

A s

ocio

-tech

nic

al syste

m How can human-generated

Web data be transformed

into a source that informs

Web system design

How to enhance Web-

based systems with

automated large-scale

human interpretation

GO

AL

S Correctness

Task Design

Experiment Methodologies

Efficiency Worker Modeling

Getting to Know the Crowd

Sustainability Incentives

Getting to Know the Crowd

Sheelagh Carpendale

Sheelagh Carpendale Canada Research Chair Information Visualization

NSERCAITFSMART Industrial Research Chair Interactive Technologies

InnoVis (Innovations in Visualization)

Interactions Lab

Computer Science University of Calgary

Interactive

Information

Visualization

I have come here to learn

bull While I do a lot of empirical research most is qualitative

bull While I think that crowd sourcing is potentially a fantastic new resource I think successful use of crowd sourcing is non-trivial

bull I hope to learn about how to work with this

Remco Chang

21 Remco Chang ndash Dagstuhl 15

From vision science to data science applying perception to problems in big data

Remco Chang

Assistant Professor Computer Science

Tufts University

22 Remco Chang ndash Dagstuhl 15

Weberrsquos Law

119889119875 = 119896119889119878

119878

Min Chen

Min Chen Professor of Scientific Visualization

University of Oxford (2011 ~ Present) Swansea University (1984 ~ 2011)

Research Interests most aspects of visualization aspects of HCI computer graphics computer vision data mining and information theory

Research Highlight

1 Volume Graphics and Visualization

2 Video Visualization

3 Visual Analytics

4 Perception and Cognition in Visualization

5 Theory of Visualization

1 2

3 4

5

Research Question in the Context of 15481

Short Version How can we use crowdsourcing to capture perceptual and cognitive measures to facilitate mathematical modelling of visualization and interaction

Long Version Information theory is perhaps the most promising mathematical framework that can underpin the fields of visualization and interaction as well as data processing and communication However to obtain its core measures various types of entropy one needs to capture a probability distribution of all states associated with a concept While such a probability distribution can be captured in a machine-centric data processing or communication pipeline it is not trivial to capture in a human-centric process Crowdsourcing offers an opportunity to conduct ldquobigrdquo observation and measurement of perceptual and cognitive activities It is thus a tool for ldquoBig Sciencerdquo

Stephan Diehl

Dagstuhl Seminar 15481 Evaluation in the Crowd Crowdsourcing and Human-Centred Experiments

Stephan Diehl Software Engineering University of Trier Germany

The Java Code Clone

Detection API

Empirical Software Engineering bull [JSS2015 ESEM2015 FSE2014 ICSM 2012

FSE 2011 ESEM 2011 hellip]

Empirical Information Visualization bull [InfoVis 2010 J of Media Psychology 410 hellip]

Analysis and Collaboration SE Tools bull [ASE2014 ASE 2012 ICSE 2011 TSE 2005 ]

Visualizing (Dynamic) Compound Graphs bull [EuroVisSTAR 2014 IV Journal 2012 InfoVis11

SOFTVIS10 EuroVis09 IV09 EuroVis08

CHASE08 AVI08GD04GD02 hellip]

Dagstuhl Seminar 15481 Evaluation in the Crowd Crowdsourcing and Human-Centred Experiments

bull How to reach your (very specific) target audience without

spamming

bull How to control for resolution animation speed color

bull What crowdsourcing platforms exist and what are their

features

Darren Edwards

Darren Edwards Swansea University

Cognitive psychology - categorization

Health and clinical psychology ndash applications of categorization on clinical populations (eg autism)

Crowdsourcing and psychological interventions

Recent validity studies confirm the use of simple psychological tasks such as decision making

Recent attempts have now explored more complex tasks such as psychological interventions (eg mindfulness)

Present studies now explore more complex forms of mindfulness

What are the limitations of using crowdsourced populations online such as when using psychological interventions or categorization

Sebastian Egger

Engineering

Technical high school

bull Telematics

electrotechnical enginieering telecommunications

bull Signal Processing for UWB systems

Sociology

Theory of human societies

Interaction amp collaboration behaviour of groups

Measurement of human behaviour

Currently Human Computer Interaction

Social amp collaborative aspects of technology use

33 26112015

Sebastian Egger

Motivational aspects of crowdworkers

CS as a collaborative work setting

Optimal Setup of CS experiments

Questions (max Nr of questions max duration)

Gamified reliability checks

Scale designs

CS vs Lab whatlsquos the ground truth

New application domains of CS

34 26112015

Interest in Crowdsourcing

Sara Fabrikant

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

humanmdashsystem (visuo-spatial) interface

spatio-temporal analytics

spatialization human navigation

multivariate spatial analysis etc)

interface design

large amp small interactive (map) displays

(ie desktop mobile VR etc)

empirical evaluations

(ie eye tracking EDA EEG etc) [httpwwwunitsevenconz]

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

human sensing observatory set up

controlled laboratory | in-situ (messy) real world setting

semi|automatic data collection of various human behavioral data streams

ie psycho-physiological perceptual cognitive and sensory-motor behaviors (through crowdsourcing )

Brian Fisher

My double research life

Brian Fisher

bull PhD in Experimental Psych

bull Postdoc w Cognitive Science

society founder amp president

bull Psychonomics Fellow

bull VIS-related symposia at Cogsci

amp APS papers on cogsci of

interaction

bull Fuzzy-logicBayes models

bull Postdoc funded by Inst for Robotics

and AI

bull VAST SC VEC VACCINE

bull Co-organized Dagstuhl ldquoInteraction

with Information for Visual Reasoningrdquo

Cogsci-based papers at VIS CHI

BELIV

My Plan

VIS

Me Psychological Science

Ujwal Gadiraju

p = 0055111

Crowd Worker

Behavior

Microtask

Design

Crowdsourcing Paradigm Quality amp Effectiveness

UJWAL GADIRAJU

Incentivization Gamification

Neha Gupta

Getting to know the crowd

Presented at Dagstuhl by

Dr David Martin amp Neha Gupta

Neha Gupta

PhD Student School of Computer Science University of Nottingham UK

Investigating crowdwork through the platform Amazon Mechanical Turk

psxng1nottinghamacuk

Neha_Gi_ji

Research amp Research teams

Research Studies

bull Qualitative study of Turkers on online forums like mturkforum and Turker Nation by Dr David Martin XRCE

bull Ethnographic Studies of Turkers in their respective homes and places of work in India by Neha Gupta UoN

Research teams

Xerox

bull David Martin Jacki OrsquoNeill (Now MSR India) Ben Hanrahan (Now PSU)

psxng1nottinghamacuk

Neha_Gi_ji

Nottingham

bull Andy Crabtree

Tom Rodden

Matthias Hirth

Matthias Hirth 48 48

About Myself

Matthias Hirth

Research Assistant at University of Wuumlrzburg

httpmatthias-hirthcom

matthiashirthinformatikuni-wuerzburgde

Crowdsourcing

Platform analysis

Platform optimization

Human factors

Use cases

Online Social Networks

Collaboration networks

Structural differences of OSNs

Application of OSN relationships

Interactive applicationsGaming

Challenges of real time interactions

Serious gamingGamification

E-Sports broadcastinglive streaming

Quality of Experience

Monitoring techniques

Quantification techniques

(Crowd-bases) test design

Social interactions

Test methodology Incentive design

Use case

Measurement technique

Identification of

qualification

Matthias Hirth 49 49

Improving Crowdsourcing Technology and the Way We Use it

Crowdsourcing technology

How to transfer well known lab procedures into an uncontrolled

crowdsourcing environment

Which crowdsourcing platform features are required to conduct

successful experiments but are not present yet

Which technical developments (can) foster the usage of

crowdsourcing instead of lab experiments

How can technological developments be used for novel incentive

mechanisms

Usage of Crowdsourcing

How can we conduct reproducible experiments in diverse

crowdsourcing environment

How can we raise the awareness of crowdsourcing employers for the

individuals behind the crowd

Tobias Hoszligfeld

The crowd community in the lab

Tobias Hoszligfeld wwwmaswiwiuni-duede

bull Background Tobias Hoszligfeld

ndash 2004-2014 Research group ldquoFuture Internet Applicationsrdquo Chair of Communication Networks Univ of Wuumlrzburg

ndash Since Sep 2014 Chair of Modeling of Adaptive Systems (head) University of Duisburg-Essen

bull Interests Application of crowdsourcing

ndash User studies on Quality of Experience (QoE) for Internet apps eg video QoE in the crowd amp in the lab methodology reliability statisticshellip [2]

ndash Mobile crowdsourcing network and context measurements eg Wifi offloading incentives privacy recommendation mechanisms etc [1]

bull Interests The crowd community

ndash Crowd = Human beings Social network wisdom of the crowd community

ndash UnderstandingUtilizing the crowd community for successful human computation For which jobs can social interaction improve the quality of work Classification

ndash How can we deploy such a system where interaction is triggered

ndash System design Task design interaction mechanisms incentives

22 Nov 2015

[1] httpdropsdagstuhldeopusvolltexte20134354

[2] httpswww3informatikuni-wuerzburgdeqoewikiqualinetcrowd

Challenges Example of Crowdsourced QoE

Tobias Hoszligfeld wwwmaswiwiuni-duede

task design

incentives

Internet

crowd

community

anonymous

remote

subjects

Internet delivery of

test data and user

response

QoE study bull testing methodology bull QoE influence factors

under investigation bull additional QoE factors bull proper subset of items bull hellip

crowdsourcing setting bull short task duration bull instructions amp training bull feedback amp questions bull hidden influence factors bull reliability of users bull hellip

heterogeneous devices

and Internet access

implementation extrinsic vs intrinsic

motivation

Processing of data

Technical aspects bull server implementation bull client application amp

compatibility bull desired test conditions bull context monitoring bull hellip

test items

lab

scientific methodologies

22 Nov 2015

Jason Jacques

Jason Jacques supervised by Per Ola Kristensson

T

Crowdsourcing Code Perception

jtj21camacuk peopledscamacukjtj21

jtjacquesgmailcom jsonjcouk

PhD Working Title

science

Crowdsourcing Interests W

hat

W

ho

Ho

w

Wh

y jtj21camacuk jsonjcouk science

Radu Jianu

Radu Jianu Assistant Professor

Florida International University (PhD Brown University lsquo12)

Interests

Applications of eye-tracking in visualization tracking what people look at rather than where they look

bull Data intensive large scale eye-tracking experiments bull Gaze-contingent visualizations bull Gaze-driven data recommendation and big data curation

Automating user studies in data visualizations (on next page)

Automating user studies

1 Software designs that can (semi)automate a broad range of quantitative evaluations by leveraging

ndash Crowdsourcing ndash Benchmark data and task taxonomies ndash Automated result interpretation

2 Moving user evaluation from the end to within the software design and implementation process as a means to choose between competitive visual designs

Christian Keimel

copy IRT

Where

What

bull Hybrid broadcast broadband technologies

bull Big Data in broadcasting environment and content-creation

bull Audio-visual QoE assessment (with crowdsourcing)

About me

Christian Keimel Slide 60

Lecturer for Digital Broadcast Engineering

Research Engineer for Media Services and Platforms

copy IRT

Crowdsourcing in large scale optimisation

The human component of Big Data

Moving crowdsourcing to lean-back devices

Crowdsourcing from the couch

Crowdsourced experiments emulating the laboratory

Is it time for new methodologies ignoring the lab

Questionshellip

Christian Keimel Slide 61

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 4: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Seminar Interests

bull Human-Centred Evaluation of Visualisations

bull How to Effectively and Correctly Use Crowdsourcing

bull Experimental methodology (between vs within)

bull When not to use crowdsourcing as well as when to use it

Benjamin Bach

Benjamin Bach Microsoft Research ndash Inria Joint Centre Saclay France

Temporal Data Dynamic Networks Domain Collaborations

Benjamin Bach Microsoft Research ndash Inria Joint Centre Saclay France

How to motivate experts in participating

What kind of feedback can we expect

Which infrastructure do we need

How to find interested domain experts

How to design (micro) tasks

Kathrin Ballweg

Kathrin Ballweg ndash About me

Interests

Computer Science

Cognitive Psychology

User Studies

PhD Focus

Perception and cognition in

network visualization

Visualization for the masses

(journalists and public ndash news readers)

231115 | Dagstuhl Seminar ndash Evaluation in the Crowd Crowdsouring and Human-Centered Experiments | K Ballweg | 9

Kathrin Ballweg

M Sc Computer Science

Technische Universitaumlt

Darmstadt

Design

Evaluation in the Crowd Crowdsourcing and

Human-Centered Experiments

Topics of Interest

Crowdsourcing Platforms vs

The Laboratory

Scientifically Rigorous Methodologies

Crowdsourcing Experiments in Human-

Computer Interaction Visualization and

Applied PerceptionGraphics

231115 | Dagstuhl Seminar ndash Evaluation in the Crowd Crowdsouring and Human-Centered Experiments | K Ballweg | 10

httpwwwtagesnetzwerkde

Rita Borgo

Introduction

Rita Borgo

Visual Computing

Swansea University

Large Data Visualisation HPC

Glyph Based Visualisation

Time Series Analysis

Perception in Visualisation

Visual Computing Group Swansea University

Prof MWJones

D Archamabault R Borgo

R S Laramee B Mora

K W Tam X Xie

My Interest in the Crowdsourcing Phenomena

bull Crowdsourcing is a precious and very appealing resource

bull Crowdsourcing is a powerful instrument

bull However it is an instrument we do not know thoroughly yet we use ithellip

Therefore

1 Can we help each other to identify ldquoguidelinesrdquo to help us use this instrument in the best way an to its full potentials

2 Is there anything WE can DO to make it even better

Alessandro Bozzon

Alessandro Bozzon

Assistant Professor TU Delft Human-Enhanced Data Management Web Information Systems Social Data Science

Faculty Fellow IBM Benelux Inclusive Enterprise

Investigator AMS Social Sensing Smart Citizens

Web Engineering Web Science

Information Retrieval

User Modelling Crowdsourcing

Human Computation

WWW

ICWE

UMAP

HT

VLDBJ

TWEB

IC

SWJ

JWE

JWS

CSCW

ISWC

Publication Venues

wwwalessandrobozzoncom

Topics of Interest

How can humans and machines better collaborate in solving

(computational) problems

Machines Social

Data

People

Process

A s

ocio

-tech

nic

al syste

m How can human-generated

Web data be transformed

into a source that informs

Web system design

How to enhance Web-

based systems with

automated large-scale

human interpretation

GO

AL

S Correctness

Task Design

Experiment Methodologies

Efficiency Worker Modeling

Getting to Know the Crowd

Sustainability Incentives

Getting to Know the Crowd

Sheelagh Carpendale

Sheelagh Carpendale Canada Research Chair Information Visualization

NSERCAITFSMART Industrial Research Chair Interactive Technologies

InnoVis (Innovations in Visualization)

Interactions Lab

Computer Science University of Calgary

Interactive

Information

Visualization

I have come here to learn

bull While I do a lot of empirical research most is qualitative

bull While I think that crowd sourcing is potentially a fantastic new resource I think successful use of crowd sourcing is non-trivial

bull I hope to learn about how to work with this

Remco Chang

21 Remco Chang ndash Dagstuhl 15

From vision science to data science applying perception to problems in big data

Remco Chang

Assistant Professor Computer Science

Tufts University

22 Remco Chang ndash Dagstuhl 15

Weberrsquos Law

119889119875 = 119896119889119878

119878

Min Chen

Min Chen Professor of Scientific Visualization

University of Oxford (2011 ~ Present) Swansea University (1984 ~ 2011)

Research Interests most aspects of visualization aspects of HCI computer graphics computer vision data mining and information theory

Research Highlight

1 Volume Graphics and Visualization

2 Video Visualization

3 Visual Analytics

4 Perception and Cognition in Visualization

5 Theory of Visualization

1 2

3 4

5

Research Question in the Context of 15481

Short Version How can we use crowdsourcing to capture perceptual and cognitive measures to facilitate mathematical modelling of visualization and interaction

Long Version Information theory is perhaps the most promising mathematical framework that can underpin the fields of visualization and interaction as well as data processing and communication However to obtain its core measures various types of entropy one needs to capture a probability distribution of all states associated with a concept While such a probability distribution can be captured in a machine-centric data processing or communication pipeline it is not trivial to capture in a human-centric process Crowdsourcing offers an opportunity to conduct ldquobigrdquo observation and measurement of perceptual and cognitive activities It is thus a tool for ldquoBig Sciencerdquo

Stephan Diehl

Dagstuhl Seminar 15481 Evaluation in the Crowd Crowdsourcing and Human-Centred Experiments

Stephan Diehl Software Engineering University of Trier Germany

The Java Code Clone

Detection API

Empirical Software Engineering bull [JSS2015 ESEM2015 FSE2014 ICSM 2012

FSE 2011 ESEM 2011 hellip]

Empirical Information Visualization bull [InfoVis 2010 J of Media Psychology 410 hellip]

Analysis and Collaboration SE Tools bull [ASE2014 ASE 2012 ICSE 2011 TSE 2005 ]

Visualizing (Dynamic) Compound Graphs bull [EuroVisSTAR 2014 IV Journal 2012 InfoVis11

SOFTVIS10 EuroVis09 IV09 EuroVis08

CHASE08 AVI08GD04GD02 hellip]

Dagstuhl Seminar 15481 Evaluation in the Crowd Crowdsourcing and Human-Centred Experiments

bull How to reach your (very specific) target audience without

spamming

bull How to control for resolution animation speed color

bull What crowdsourcing platforms exist and what are their

features

Darren Edwards

Darren Edwards Swansea University

Cognitive psychology - categorization

Health and clinical psychology ndash applications of categorization on clinical populations (eg autism)

Crowdsourcing and psychological interventions

Recent validity studies confirm the use of simple psychological tasks such as decision making

Recent attempts have now explored more complex tasks such as psychological interventions (eg mindfulness)

Present studies now explore more complex forms of mindfulness

What are the limitations of using crowdsourced populations online such as when using psychological interventions or categorization

Sebastian Egger

Engineering

Technical high school

bull Telematics

electrotechnical enginieering telecommunications

bull Signal Processing for UWB systems

Sociology

Theory of human societies

Interaction amp collaboration behaviour of groups

Measurement of human behaviour

Currently Human Computer Interaction

Social amp collaborative aspects of technology use

33 26112015

Sebastian Egger

Motivational aspects of crowdworkers

CS as a collaborative work setting

Optimal Setup of CS experiments

Questions (max Nr of questions max duration)

Gamified reliability checks

Scale designs

CS vs Lab whatlsquos the ground truth

New application domains of CS

34 26112015

Interest in Crowdsourcing

Sara Fabrikant

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

humanmdashsystem (visuo-spatial) interface

spatio-temporal analytics

spatialization human navigation

multivariate spatial analysis etc)

interface design

large amp small interactive (map) displays

(ie desktop mobile VR etc)

empirical evaluations

(ie eye tracking EDA EEG etc) [httpwwwunitsevenconz]

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

human sensing observatory set up

controlled laboratory | in-situ (messy) real world setting

semi|automatic data collection of various human behavioral data streams

ie psycho-physiological perceptual cognitive and sensory-motor behaviors (through crowdsourcing )

Brian Fisher

My double research life

Brian Fisher

bull PhD in Experimental Psych

bull Postdoc w Cognitive Science

society founder amp president

bull Psychonomics Fellow

bull VIS-related symposia at Cogsci

amp APS papers on cogsci of

interaction

bull Fuzzy-logicBayes models

bull Postdoc funded by Inst for Robotics

and AI

bull VAST SC VEC VACCINE

bull Co-organized Dagstuhl ldquoInteraction

with Information for Visual Reasoningrdquo

Cogsci-based papers at VIS CHI

BELIV

My Plan

VIS

Me Psychological Science

Ujwal Gadiraju

p = 0055111

Crowd Worker

Behavior

Microtask

Design

Crowdsourcing Paradigm Quality amp Effectiveness

UJWAL GADIRAJU

Incentivization Gamification

Neha Gupta

Getting to know the crowd

Presented at Dagstuhl by

Dr David Martin amp Neha Gupta

Neha Gupta

PhD Student School of Computer Science University of Nottingham UK

Investigating crowdwork through the platform Amazon Mechanical Turk

psxng1nottinghamacuk

Neha_Gi_ji

Research amp Research teams

Research Studies

bull Qualitative study of Turkers on online forums like mturkforum and Turker Nation by Dr David Martin XRCE

bull Ethnographic Studies of Turkers in their respective homes and places of work in India by Neha Gupta UoN

Research teams

Xerox

bull David Martin Jacki OrsquoNeill (Now MSR India) Ben Hanrahan (Now PSU)

psxng1nottinghamacuk

Neha_Gi_ji

Nottingham

bull Andy Crabtree

Tom Rodden

Matthias Hirth

Matthias Hirth 48 48

About Myself

Matthias Hirth

Research Assistant at University of Wuumlrzburg

httpmatthias-hirthcom

matthiashirthinformatikuni-wuerzburgde

Crowdsourcing

Platform analysis

Platform optimization

Human factors

Use cases

Online Social Networks

Collaboration networks

Structural differences of OSNs

Application of OSN relationships

Interactive applicationsGaming

Challenges of real time interactions

Serious gamingGamification

E-Sports broadcastinglive streaming

Quality of Experience

Monitoring techniques

Quantification techniques

(Crowd-bases) test design

Social interactions

Test methodology Incentive design

Use case

Measurement technique

Identification of

qualification

Matthias Hirth 49 49

Improving Crowdsourcing Technology and the Way We Use it

Crowdsourcing technology

How to transfer well known lab procedures into an uncontrolled

crowdsourcing environment

Which crowdsourcing platform features are required to conduct

successful experiments but are not present yet

Which technical developments (can) foster the usage of

crowdsourcing instead of lab experiments

How can technological developments be used for novel incentive

mechanisms

Usage of Crowdsourcing

How can we conduct reproducible experiments in diverse

crowdsourcing environment

How can we raise the awareness of crowdsourcing employers for the

individuals behind the crowd

Tobias Hoszligfeld

The crowd community in the lab

Tobias Hoszligfeld wwwmaswiwiuni-duede

bull Background Tobias Hoszligfeld

ndash 2004-2014 Research group ldquoFuture Internet Applicationsrdquo Chair of Communication Networks Univ of Wuumlrzburg

ndash Since Sep 2014 Chair of Modeling of Adaptive Systems (head) University of Duisburg-Essen

bull Interests Application of crowdsourcing

ndash User studies on Quality of Experience (QoE) for Internet apps eg video QoE in the crowd amp in the lab methodology reliability statisticshellip [2]

ndash Mobile crowdsourcing network and context measurements eg Wifi offloading incentives privacy recommendation mechanisms etc [1]

bull Interests The crowd community

ndash Crowd = Human beings Social network wisdom of the crowd community

ndash UnderstandingUtilizing the crowd community for successful human computation For which jobs can social interaction improve the quality of work Classification

ndash How can we deploy such a system where interaction is triggered

ndash System design Task design interaction mechanisms incentives

22 Nov 2015

[1] httpdropsdagstuhldeopusvolltexte20134354

[2] httpswww3informatikuni-wuerzburgdeqoewikiqualinetcrowd

Challenges Example of Crowdsourced QoE

Tobias Hoszligfeld wwwmaswiwiuni-duede

task design

incentives

Internet

crowd

community

anonymous

remote

subjects

Internet delivery of

test data and user

response

QoE study bull testing methodology bull QoE influence factors

under investigation bull additional QoE factors bull proper subset of items bull hellip

crowdsourcing setting bull short task duration bull instructions amp training bull feedback amp questions bull hidden influence factors bull reliability of users bull hellip

heterogeneous devices

and Internet access

implementation extrinsic vs intrinsic

motivation

Processing of data

Technical aspects bull server implementation bull client application amp

compatibility bull desired test conditions bull context monitoring bull hellip

test items

lab

scientific methodologies

22 Nov 2015

Jason Jacques

Jason Jacques supervised by Per Ola Kristensson

T

Crowdsourcing Code Perception

jtj21camacuk peopledscamacukjtj21

jtjacquesgmailcom jsonjcouk

PhD Working Title

science

Crowdsourcing Interests W

hat

W

ho

Ho

w

Wh

y jtj21camacuk jsonjcouk science

Radu Jianu

Radu Jianu Assistant Professor

Florida International University (PhD Brown University lsquo12)

Interests

Applications of eye-tracking in visualization tracking what people look at rather than where they look

bull Data intensive large scale eye-tracking experiments bull Gaze-contingent visualizations bull Gaze-driven data recommendation and big data curation

Automating user studies in data visualizations (on next page)

Automating user studies

1 Software designs that can (semi)automate a broad range of quantitative evaluations by leveraging

ndash Crowdsourcing ndash Benchmark data and task taxonomies ndash Automated result interpretation

2 Moving user evaluation from the end to within the software design and implementation process as a means to choose between competitive visual designs

Christian Keimel

copy IRT

Where

What

bull Hybrid broadcast broadband technologies

bull Big Data in broadcasting environment and content-creation

bull Audio-visual QoE assessment (with crowdsourcing)

About me

Christian Keimel Slide 60

Lecturer for Digital Broadcast Engineering

Research Engineer for Media Services and Platforms

copy IRT

Crowdsourcing in large scale optimisation

The human component of Big Data

Moving crowdsourcing to lean-back devices

Crowdsourcing from the couch

Crowdsourced experiments emulating the laboratory

Is it time for new methodologies ignoring the lab

Questionshellip

Christian Keimel Slide 61

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 5: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Benjamin Bach

Benjamin Bach Microsoft Research ndash Inria Joint Centre Saclay France

Temporal Data Dynamic Networks Domain Collaborations

Benjamin Bach Microsoft Research ndash Inria Joint Centre Saclay France

How to motivate experts in participating

What kind of feedback can we expect

Which infrastructure do we need

How to find interested domain experts

How to design (micro) tasks

Kathrin Ballweg

Kathrin Ballweg ndash About me

Interests

Computer Science

Cognitive Psychology

User Studies

PhD Focus

Perception and cognition in

network visualization

Visualization for the masses

(journalists and public ndash news readers)

231115 | Dagstuhl Seminar ndash Evaluation in the Crowd Crowdsouring and Human-Centered Experiments | K Ballweg | 9

Kathrin Ballweg

M Sc Computer Science

Technische Universitaumlt

Darmstadt

Design

Evaluation in the Crowd Crowdsourcing and

Human-Centered Experiments

Topics of Interest

Crowdsourcing Platforms vs

The Laboratory

Scientifically Rigorous Methodologies

Crowdsourcing Experiments in Human-

Computer Interaction Visualization and

Applied PerceptionGraphics

231115 | Dagstuhl Seminar ndash Evaluation in the Crowd Crowdsouring and Human-Centered Experiments | K Ballweg | 10

httpwwwtagesnetzwerkde

Rita Borgo

Introduction

Rita Borgo

Visual Computing

Swansea University

Large Data Visualisation HPC

Glyph Based Visualisation

Time Series Analysis

Perception in Visualisation

Visual Computing Group Swansea University

Prof MWJones

D Archamabault R Borgo

R S Laramee B Mora

K W Tam X Xie

My Interest in the Crowdsourcing Phenomena

bull Crowdsourcing is a precious and very appealing resource

bull Crowdsourcing is a powerful instrument

bull However it is an instrument we do not know thoroughly yet we use ithellip

Therefore

1 Can we help each other to identify ldquoguidelinesrdquo to help us use this instrument in the best way an to its full potentials

2 Is there anything WE can DO to make it even better

Alessandro Bozzon

Alessandro Bozzon

Assistant Professor TU Delft Human-Enhanced Data Management Web Information Systems Social Data Science

Faculty Fellow IBM Benelux Inclusive Enterprise

Investigator AMS Social Sensing Smart Citizens

Web Engineering Web Science

Information Retrieval

User Modelling Crowdsourcing

Human Computation

WWW

ICWE

UMAP

HT

VLDBJ

TWEB

IC

SWJ

JWE

JWS

CSCW

ISWC

Publication Venues

wwwalessandrobozzoncom

Topics of Interest

How can humans and machines better collaborate in solving

(computational) problems

Machines Social

Data

People

Process

A s

ocio

-tech

nic

al syste

m How can human-generated

Web data be transformed

into a source that informs

Web system design

How to enhance Web-

based systems with

automated large-scale

human interpretation

GO

AL

S Correctness

Task Design

Experiment Methodologies

Efficiency Worker Modeling

Getting to Know the Crowd

Sustainability Incentives

Getting to Know the Crowd

Sheelagh Carpendale

Sheelagh Carpendale Canada Research Chair Information Visualization

NSERCAITFSMART Industrial Research Chair Interactive Technologies

InnoVis (Innovations in Visualization)

Interactions Lab

Computer Science University of Calgary

Interactive

Information

Visualization

I have come here to learn

bull While I do a lot of empirical research most is qualitative

bull While I think that crowd sourcing is potentially a fantastic new resource I think successful use of crowd sourcing is non-trivial

bull I hope to learn about how to work with this

Remco Chang

21 Remco Chang ndash Dagstuhl 15

From vision science to data science applying perception to problems in big data

Remco Chang

Assistant Professor Computer Science

Tufts University

22 Remco Chang ndash Dagstuhl 15

Weberrsquos Law

119889119875 = 119896119889119878

119878

Min Chen

Min Chen Professor of Scientific Visualization

University of Oxford (2011 ~ Present) Swansea University (1984 ~ 2011)

Research Interests most aspects of visualization aspects of HCI computer graphics computer vision data mining and information theory

Research Highlight

1 Volume Graphics and Visualization

2 Video Visualization

3 Visual Analytics

4 Perception and Cognition in Visualization

5 Theory of Visualization

1 2

3 4

5

Research Question in the Context of 15481

Short Version How can we use crowdsourcing to capture perceptual and cognitive measures to facilitate mathematical modelling of visualization and interaction

Long Version Information theory is perhaps the most promising mathematical framework that can underpin the fields of visualization and interaction as well as data processing and communication However to obtain its core measures various types of entropy one needs to capture a probability distribution of all states associated with a concept While such a probability distribution can be captured in a machine-centric data processing or communication pipeline it is not trivial to capture in a human-centric process Crowdsourcing offers an opportunity to conduct ldquobigrdquo observation and measurement of perceptual and cognitive activities It is thus a tool for ldquoBig Sciencerdquo

Stephan Diehl

Dagstuhl Seminar 15481 Evaluation in the Crowd Crowdsourcing and Human-Centred Experiments

Stephan Diehl Software Engineering University of Trier Germany

The Java Code Clone

Detection API

Empirical Software Engineering bull [JSS2015 ESEM2015 FSE2014 ICSM 2012

FSE 2011 ESEM 2011 hellip]

Empirical Information Visualization bull [InfoVis 2010 J of Media Psychology 410 hellip]

Analysis and Collaboration SE Tools bull [ASE2014 ASE 2012 ICSE 2011 TSE 2005 ]

Visualizing (Dynamic) Compound Graphs bull [EuroVisSTAR 2014 IV Journal 2012 InfoVis11

SOFTVIS10 EuroVis09 IV09 EuroVis08

CHASE08 AVI08GD04GD02 hellip]

Dagstuhl Seminar 15481 Evaluation in the Crowd Crowdsourcing and Human-Centred Experiments

bull How to reach your (very specific) target audience without

spamming

bull How to control for resolution animation speed color

bull What crowdsourcing platforms exist and what are their

features

Darren Edwards

Darren Edwards Swansea University

Cognitive psychology - categorization

Health and clinical psychology ndash applications of categorization on clinical populations (eg autism)

Crowdsourcing and psychological interventions

Recent validity studies confirm the use of simple psychological tasks such as decision making

Recent attempts have now explored more complex tasks such as psychological interventions (eg mindfulness)

Present studies now explore more complex forms of mindfulness

What are the limitations of using crowdsourced populations online such as when using psychological interventions or categorization

Sebastian Egger

Engineering

Technical high school

bull Telematics

electrotechnical enginieering telecommunications

bull Signal Processing for UWB systems

Sociology

Theory of human societies

Interaction amp collaboration behaviour of groups

Measurement of human behaviour

Currently Human Computer Interaction

Social amp collaborative aspects of technology use

33 26112015

Sebastian Egger

Motivational aspects of crowdworkers

CS as a collaborative work setting

Optimal Setup of CS experiments

Questions (max Nr of questions max duration)

Gamified reliability checks

Scale designs

CS vs Lab whatlsquos the ground truth

New application domains of CS

34 26112015

Interest in Crowdsourcing

Sara Fabrikant

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

humanmdashsystem (visuo-spatial) interface

spatio-temporal analytics

spatialization human navigation

multivariate spatial analysis etc)

interface design

large amp small interactive (map) displays

(ie desktop mobile VR etc)

empirical evaluations

(ie eye tracking EDA EEG etc) [httpwwwunitsevenconz]

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

human sensing observatory set up

controlled laboratory | in-situ (messy) real world setting

semi|automatic data collection of various human behavioral data streams

ie psycho-physiological perceptual cognitive and sensory-motor behaviors (through crowdsourcing )

Brian Fisher

My double research life

Brian Fisher

bull PhD in Experimental Psych

bull Postdoc w Cognitive Science

society founder amp president

bull Psychonomics Fellow

bull VIS-related symposia at Cogsci

amp APS papers on cogsci of

interaction

bull Fuzzy-logicBayes models

bull Postdoc funded by Inst for Robotics

and AI

bull VAST SC VEC VACCINE

bull Co-organized Dagstuhl ldquoInteraction

with Information for Visual Reasoningrdquo

Cogsci-based papers at VIS CHI

BELIV

My Plan

VIS

Me Psychological Science

Ujwal Gadiraju

p = 0055111

Crowd Worker

Behavior

Microtask

Design

Crowdsourcing Paradigm Quality amp Effectiveness

UJWAL GADIRAJU

Incentivization Gamification

Neha Gupta

Getting to know the crowd

Presented at Dagstuhl by

Dr David Martin amp Neha Gupta

Neha Gupta

PhD Student School of Computer Science University of Nottingham UK

Investigating crowdwork through the platform Amazon Mechanical Turk

psxng1nottinghamacuk

Neha_Gi_ji

Research amp Research teams

Research Studies

bull Qualitative study of Turkers on online forums like mturkforum and Turker Nation by Dr David Martin XRCE

bull Ethnographic Studies of Turkers in their respective homes and places of work in India by Neha Gupta UoN

Research teams

Xerox

bull David Martin Jacki OrsquoNeill (Now MSR India) Ben Hanrahan (Now PSU)

psxng1nottinghamacuk

Neha_Gi_ji

Nottingham

bull Andy Crabtree

Tom Rodden

Matthias Hirth

Matthias Hirth 48 48

About Myself

Matthias Hirth

Research Assistant at University of Wuumlrzburg

httpmatthias-hirthcom

matthiashirthinformatikuni-wuerzburgde

Crowdsourcing

Platform analysis

Platform optimization

Human factors

Use cases

Online Social Networks

Collaboration networks

Structural differences of OSNs

Application of OSN relationships

Interactive applicationsGaming

Challenges of real time interactions

Serious gamingGamification

E-Sports broadcastinglive streaming

Quality of Experience

Monitoring techniques

Quantification techniques

(Crowd-bases) test design

Social interactions

Test methodology Incentive design

Use case

Measurement technique

Identification of

qualification

Matthias Hirth 49 49

Improving Crowdsourcing Technology and the Way We Use it

Crowdsourcing technology

How to transfer well known lab procedures into an uncontrolled

crowdsourcing environment

Which crowdsourcing platform features are required to conduct

successful experiments but are not present yet

Which technical developments (can) foster the usage of

crowdsourcing instead of lab experiments

How can technological developments be used for novel incentive

mechanisms

Usage of Crowdsourcing

How can we conduct reproducible experiments in diverse

crowdsourcing environment

How can we raise the awareness of crowdsourcing employers for the

individuals behind the crowd

Tobias Hoszligfeld

The crowd community in the lab

Tobias Hoszligfeld wwwmaswiwiuni-duede

bull Background Tobias Hoszligfeld

ndash 2004-2014 Research group ldquoFuture Internet Applicationsrdquo Chair of Communication Networks Univ of Wuumlrzburg

ndash Since Sep 2014 Chair of Modeling of Adaptive Systems (head) University of Duisburg-Essen

bull Interests Application of crowdsourcing

ndash User studies on Quality of Experience (QoE) for Internet apps eg video QoE in the crowd amp in the lab methodology reliability statisticshellip [2]

ndash Mobile crowdsourcing network and context measurements eg Wifi offloading incentives privacy recommendation mechanisms etc [1]

bull Interests The crowd community

ndash Crowd = Human beings Social network wisdom of the crowd community

ndash UnderstandingUtilizing the crowd community for successful human computation For which jobs can social interaction improve the quality of work Classification

ndash How can we deploy such a system where interaction is triggered

ndash System design Task design interaction mechanisms incentives

22 Nov 2015

[1] httpdropsdagstuhldeopusvolltexte20134354

[2] httpswww3informatikuni-wuerzburgdeqoewikiqualinetcrowd

Challenges Example of Crowdsourced QoE

Tobias Hoszligfeld wwwmaswiwiuni-duede

task design

incentives

Internet

crowd

community

anonymous

remote

subjects

Internet delivery of

test data and user

response

QoE study bull testing methodology bull QoE influence factors

under investigation bull additional QoE factors bull proper subset of items bull hellip

crowdsourcing setting bull short task duration bull instructions amp training bull feedback amp questions bull hidden influence factors bull reliability of users bull hellip

heterogeneous devices

and Internet access

implementation extrinsic vs intrinsic

motivation

Processing of data

Technical aspects bull server implementation bull client application amp

compatibility bull desired test conditions bull context monitoring bull hellip

test items

lab

scientific methodologies

22 Nov 2015

Jason Jacques

Jason Jacques supervised by Per Ola Kristensson

T

Crowdsourcing Code Perception

jtj21camacuk peopledscamacukjtj21

jtjacquesgmailcom jsonjcouk

PhD Working Title

science

Crowdsourcing Interests W

hat

W

ho

Ho

w

Wh

y jtj21camacuk jsonjcouk science

Radu Jianu

Radu Jianu Assistant Professor

Florida International University (PhD Brown University lsquo12)

Interests

Applications of eye-tracking in visualization tracking what people look at rather than where they look

bull Data intensive large scale eye-tracking experiments bull Gaze-contingent visualizations bull Gaze-driven data recommendation and big data curation

Automating user studies in data visualizations (on next page)

Automating user studies

1 Software designs that can (semi)automate a broad range of quantitative evaluations by leveraging

ndash Crowdsourcing ndash Benchmark data and task taxonomies ndash Automated result interpretation

2 Moving user evaluation from the end to within the software design and implementation process as a means to choose between competitive visual designs

Christian Keimel

copy IRT

Where

What

bull Hybrid broadcast broadband technologies

bull Big Data in broadcasting environment and content-creation

bull Audio-visual QoE assessment (with crowdsourcing)

About me

Christian Keimel Slide 60

Lecturer for Digital Broadcast Engineering

Research Engineer for Media Services and Platforms

copy IRT

Crowdsourcing in large scale optimisation

The human component of Big Data

Moving crowdsourcing to lean-back devices

Crowdsourcing from the couch

Crowdsourced experiments emulating the laboratory

Is it time for new methodologies ignoring the lab

Questionshellip

Christian Keimel Slide 61

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 6: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Benjamin Bach Microsoft Research ndash Inria Joint Centre Saclay France

Temporal Data Dynamic Networks Domain Collaborations

Benjamin Bach Microsoft Research ndash Inria Joint Centre Saclay France

How to motivate experts in participating

What kind of feedback can we expect

Which infrastructure do we need

How to find interested domain experts

How to design (micro) tasks

Kathrin Ballweg

Kathrin Ballweg ndash About me

Interests

Computer Science

Cognitive Psychology

User Studies

PhD Focus

Perception and cognition in

network visualization

Visualization for the masses

(journalists and public ndash news readers)

231115 | Dagstuhl Seminar ndash Evaluation in the Crowd Crowdsouring and Human-Centered Experiments | K Ballweg | 9

Kathrin Ballweg

M Sc Computer Science

Technische Universitaumlt

Darmstadt

Design

Evaluation in the Crowd Crowdsourcing and

Human-Centered Experiments

Topics of Interest

Crowdsourcing Platforms vs

The Laboratory

Scientifically Rigorous Methodologies

Crowdsourcing Experiments in Human-

Computer Interaction Visualization and

Applied PerceptionGraphics

231115 | Dagstuhl Seminar ndash Evaluation in the Crowd Crowdsouring and Human-Centered Experiments | K Ballweg | 10

httpwwwtagesnetzwerkde

Rita Borgo

Introduction

Rita Borgo

Visual Computing

Swansea University

Large Data Visualisation HPC

Glyph Based Visualisation

Time Series Analysis

Perception in Visualisation

Visual Computing Group Swansea University

Prof MWJones

D Archamabault R Borgo

R S Laramee B Mora

K W Tam X Xie

My Interest in the Crowdsourcing Phenomena

bull Crowdsourcing is a precious and very appealing resource

bull Crowdsourcing is a powerful instrument

bull However it is an instrument we do not know thoroughly yet we use ithellip

Therefore

1 Can we help each other to identify ldquoguidelinesrdquo to help us use this instrument in the best way an to its full potentials

2 Is there anything WE can DO to make it even better

Alessandro Bozzon

Alessandro Bozzon

Assistant Professor TU Delft Human-Enhanced Data Management Web Information Systems Social Data Science

Faculty Fellow IBM Benelux Inclusive Enterprise

Investigator AMS Social Sensing Smart Citizens

Web Engineering Web Science

Information Retrieval

User Modelling Crowdsourcing

Human Computation

WWW

ICWE

UMAP

HT

VLDBJ

TWEB

IC

SWJ

JWE

JWS

CSCW

ISWC

Publication Venues

wwwalessandrobozzoncom

Topics of Interest

How can humans and machines better collaborate in solving

(computational) problems

Machines Social

Data

People

Process

A s

ocio

-tech

nic

al syste

m How can human-generated

Web data be transformed

into a source that informs

Web system design

How to enhance Web-

based systems with

automated large-scale

human interpretation

GO

AL

S Correctness

Task Design

Experiment Methodologies

Efficiency Worker Modeling

Getting to Know the Crowd

Sustainability Incentives

Getting to Know the Crowd

Sheelagh Carpendale

Sheelagh Carpendale Canada Research Chair Information Visualization

NSERCAITFSMART Industrial Research Chair Interactive Technologies

InnoVis (Innovations in Visualization)

Interactions Lab

Computer Science University of Calgary

Interactive

Information

Visualization

I have come here to learn

bull While I do a lot of empirical research most is qualitative

bull While I think that crowd sourcing is potentially a fantastic new resource I think successful use of crowd sourcing is non-trivial

bull I hope to learn about how to work with this

Remco Chang

21 Remco Chang ndash Dagstuhl 15

From vision science to data science applying perception to problems in big data

Remco Chang

Assistant Professor Computer Science

Tufts University

22 Remco Chang ndash Dagstuhl 15

Weberrsquos Law

119889119875 = 119896119889119878

119878

Min Chen

Min Chen Professor of Scientific Visualization

University of Oxford (2011 ~ Present) Swansea University (1984 ~ 2011)

Research Interests most aspects of visualization aspects of HCI computer graphics computer vision data mining and information theory

Research Highlight

1 Volume Graphics and Visualization

2 Video Visualization

3 Visual Analytics

4 Perception and Cognition in Visualization

5 Theory of Visualization

1 2

3 4

5

Research Question in the Context of 15481

Short Version How can we use crowdsourcing to capture perceptual and cognitive measures to facilitate mathematical modelling of visualization and interaction

Long Version Information theory is perhaps the most promising mathematical framework that can underpin the fields of visualization and interaction as well as data processing and communication However to obtain its core measures various types of entropy one needs to capture a probability distribution of all states associated with a concept While such a probability distribution can be captured in a machine-centric data processing or communication pipeline it is not trivial to capture in a human-centric process Crowdsourcing offers an opportunity to conduct ldquobigrdquo observation and measurement of perceptual and cognitive activities It is thus a tool for ldquoBig Sciencerdquo

Stephan Diehl

Dagstuhl Seminar 15481 Evaluation in the Crowd Crowdsourcing and Human-Centred Experiments

Stephan Diehl Software Engineering University of Trier Germany

The Java Code Clone

Detection API

Empirical Software Engineering bull [JSS2015 ESEM2015 FSE2014 ICSM 2012

FSE 2011 ESEM 2011 hellip]

Empirical Information Visualization bull [InfoVis 2010 J of Media Psychology 410 hellip]

Analysis and Collaboration SE Tools bull [ASE2014 ASE 2012 ICSE 2011 TSE 2005 ]

Visualizing (Dynamic) Compound Graphs bull [EuroVisSTAR 2014 IV Journal 2012 InfoVis11

SOFTVIS10 EuroVis09 IV09 EuroVis08

CHASE08 AVI08GD04GD02 hellip]

Dagstuhl Seminar 15481 Evaluation in the Crowd Crowdsourcing and Human-Centred Experiments

bull How to reach your (very specific) target audience without

spamming

bull How to control for resolution animation speed color

bull What crowdsourcing platforms exist and what are their

features

Darren Edwards

Darren Edwards Swansea University

Cognitive psychology - categorization

Health and clinical psychology ndash applications of categorization on clinical populations (eg autism)

Crowdsourcing and psychological interventions

Recent validity studies confirm the use of simple psychological tasks such as decision making

Recent attempts have now explored more complex tasks such as psychological interventions (eg mindfulness)

Present studies now explore more complex forms of mindfulness

What are the limitations of using crowdsourced populations online such as when using psychological interventions or categorization

Sebastian Egger

Engineering

Technical high school

bull Telematics

electrotechnical enginieering telecommunications

bull Signal Processing for UWB systems

Sociology

Theory of human societies

Interaction amp collaboration behaviour of groups

Measurement of human behaviour

Currently Human Computer Interaction

Social amp collaborative aspects of technology use

33 26112015

Sebastian Egger

Motivational aspects of crowdworkers

CS as a collaborative work setting

Optimal Setup of CS experiments

Questions (max Nr of questions max duration)

Gamified reliability checks

Scale designs

CS vs Lab whatlsquos the ground truth

New application domains of CS

34 26112015

Interest in Crowdsourcing

Sara Fabrikant

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

humanmdashsystem (visuo-spatial) interface

spatio-temporal analytics

spatialization human navigation

multivariate spatial analysis etc)

interface design

large amp small interactive (map) displays

(ie desktop mobile VR etc)

empirical evaluations

(ie eye tracking EDA EEG etc) [httpwwwunitsevenconz]

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

human sensing observatory set up

controlled laboratory | in-situ (messy) real world setting

semi|automatic data collection of various human behavioral data streams

ie psycho-physiological perceptual cognitive and sensory-motor behaviors (through crowdsourcing )

Brian Fisher

My double research life

Brian Fisher

bull PhD in Experimental Psych

bull Postdoc w Cognitive Science

society founder amp president

bull Psychonomics Fellow

bull VIS-related symposia at Cogsci

amp APS papers on cogsci of

interaction

bull Fuzzy-logicBayes models

bull Postdoc funded by Inst for Robotics

and AI

bull VAST SC VEC VACCINE

bull Co-organized Dagstuhl ldquoInteraction

with Information for Visual Reasoningrdquo

Cogsci-based papers at VIS CHI

BELIV

My Plan

VIS

Me Psychological Science

Ujwal Gadiraju

p = 0055111

Crowd Worker

Behavior

Microtask

Design

Crowdsourcing Paradigm Quality amp Effectiveness

UJWAL GADIRAJU

Incentivization Gamification

Neha Gupta

Getting to know the crowd

Presented at Dagstuhl by

Dr David Martin amp Neha Gupta

Neha Gupta

PhD Student School of Computer Science University of Nottingham UK

Investigating crowdwork through the platform Amazon Mechanical Turk

psxng1nottinghamacuk

Neha_Gi_ji

Research amp Research teams

Research Studies

bull Qualitative study of Turkers on online forums like mturkforum and Turker Nation by Dr David Martin XRCE

bull Ethnographic Studies of Turkers in their respective homes and places of work in India by Neha Gupta UoN

Research teams

Xerox

bull David Martin Jacki OrsquoNeill (Now MSR India) Ben Hanrahan (Now PSU)

psxng1nottinghamacuk

Neha_Gi_ji

Nottingham

bull Andy Crabtree

Tom Rodden

Matthias Hirth

Matthias Hirth 48 48

About Myself

Matthias Hirth

Research Assistant at University of Wuumlrzburg

httpmatthias-hirthcom

matthiashirthinformatikuni-wuerzburgde

Crowdsourcing

Platform analysis

Platform optimization

Human factors

Use cases

Online Social Networks

Collaboration networks

Structural differences of OSNs

Application of OSN relationships

Interactive applicationsGaming

Challenges of real time interactions

Serious gamingGamification

E-Sports broadcastinglive streaming

Quality of Experience

Monitoring techniques

Quantification techniques

(Crowd-bases) test design

Social interactions

Test methodology Incentive design

Use case

Measurement technique

Identification of

qualification

Matthias Hirth 49 49

Improving Crowdsourcing Technology and the Way We Use it

Crowdsourcing technology

How to transfer well known lab procedures into an uncontrolled

crowdsourcing environment

Which crowdsourcing platform features are required to conduct

successful experiments but are not present yet

Which technical developments (can) foster the usage of

crowdsourcing instead of lab experiments

How can technological developments be used for novel incentive

mechanisms

Usage of Crowdsourcing

How can we conduct reproducible experiments in diverse

crowdsourcing environment

How can we raise the awareness of crowdsourcing employers for the

individuals behind the crowd

Tobias Hoszligfeld

The crowd community in the lab

Tobias Hoszligfeld wwwmaswiwiuni-duede

bull Background Tobias Hoszligfeld

ndash 2004-2014 Research group ldquoFuture Internet Applicationsrdquo Chair of Communication Networks Univ of Wuumlrzburg

ndash Since Sep 2014 Chair of Modeling of Adaptive Systems (head) University of Duisburg-Essen

bull Interests Application of crowdsourcing

ndash User studies on Quality of Experience (QoE) for Internet apps eg video QoE in the crowd amp in the lab methodology reliability statisticshellip [2]

ndash Mobile crowdsourcing network and context measurements eg Wifi offloading incentives privacy recommendation mechanisms etc [1]

bull Interests The crowd community

ndash Crowd = Human beings Social network wisdom of the crowd community

ndash UnderstandingUtilizing the crowd community for successful human computation For which jobs can social interaction improve the quality of work Classification

ndash How can we deploy such a system where interaction is triggered

ndash System design Task design interaction mechanisms incentives

22 Nov 2015

[1] httpdropsdagstuhldeopusvolltexte20134354

[2] httpswww3informatikuni-wuerzburgdeqoewikiqualinetcrowd

Challenges Example of Crowdsourced QoE

Tobias Hoszligfeld wwwmaswiwiuni-duede

task design

incentives

Internet

crowd

community

anonymous

remote

subjects

Internet delivery of

test data and user

response

QoE study bull testing methodology bull QoE influence factors

under investigation bull additional QoE factors bull proper subset of items bull hellip

crowdsourcing setting bull short task duration bull instructions amp training bull feedback amp questions bull hidden influence factors bull reliability of users bull hellip

heterogeneous devices

and Internet access

implementation extrinsic vs intrinsic

motivation

Processing of data

Technical aspects bull server implementation bull client application amp

compatibility bull desired test conditions bull context monitoring bull hellip

test items

lab

scientific methodologies

22 Nov 2015

Jason Jacques

Jason Jacques supervised by Per Ola Kristensson

T

Crowdsourcing Code Perception

jtj21camacuk peopledscamacukjtj21

jtjacquesgmailcom jsonjcouk

PhD Working Title

science

Crowdsourcing Interests W

hat

W

ho

Ho

w

Wh

y jtj21camacuk jsonjcouk science

Radu Jianu

Radu Jianu Assistant Professor

Florida International University (PhD Brown University lsquo12)

Interests

Applications of eye-tracking in visualization tracking what people look at rather than where they look

bull Data intensive large scale eye-tracking experiments bull Gaze-contingent visualizations bull Gaze-driven data recommendation and big data curation

Automating user studies in data visualizations (on next page)

Automating user studies

1 Software designs that can (semi)automate a broad range of quantitative evaluations by leveraging

ndash Crowdsourcing ndash Benchmark data and task taxonomies ndash Automated result interpretation

2 Moving user evaluation from the end to within the software design and implementation process as a means to choose between competitive visual designs

Christian Keimel

copy IRT

Where

What

bull Hybrid broadcast broadband technologies

bull Big Data in broadcasting environment and content-creation

bull Audio-visual QoE assessment (with crowdsourcing)

About me

Christian Keimel Slide 60

Lecturer for Digital Broadcast Engineering

Research Engineer for Media Services and Platforms

copy IRT

Crowdsourcing in large scale optimisation

The human component of Big Data

Moving crowdsourcing to lean-back devices

Crowdsourcing from the couch

Crowdsourced experiments emulating the laboratory

Is it time for new methodologies ignoring the lab

Questionshellip

Christian Keimel Slide 61

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 7: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Benjamin Bach Microsoft Research ndash Inria Joint Centre Saclay France

How to motivate experts in participating

What kind of feedback can we expect

Which infrastructure do we need

How to find interested domain experts

How to design (micro) tasks

Kathrin Ballweg

Kathrin Ballweg ndash About me

Interests

Computer Science

Cognitive Psychology

User Studies

PhD Focus

Perception and cognition in

network visualization

Visualization for the masses

(journalists and public ndash news readers)

231115 | Dagstuhl Seminar ndash Evaluation in the Crowd Crowdsouring and Human-Centered Experiments | K Ballweg | 9

Kathrin Ballweg

M Sc Computer Science

Technische Universitaumlt

Darmstadt

Design

Evaluation in the Crowd Crowdsourcing and

Human-Centered Experiments

Topics of Interest

Crowdsourcing Platforms vs

The Laboratory

Scientifically Rigorous Methodologies

Crowdsourcing Experiments in Human-

Computer Interaction Visualization and

Applied PerceptionGraphics

231115 | Dagstuhl Seminar ndash Evaluation in the Crowd Crowdsouring and Human-Centered Experiments | K Ballweg | 10

httpwwwtagesnetzwerkde

Rita Borgo

Introduction

Rita Borgo

Visual Computing

Swansea University

Large Data Visualisation HPC

Glyph Based Visualisation

Time Series Analysis

Perception in Visualisation

Visual Computing Group Swansea University

Prof MWJones

D Archamabault R Borgo

R S Laramee B Mora

K W Tam X Xie

My Interest in the Crowdsourcing Phenomena

bull Crowdsourcing is a precious and very appealing resource

bull Crowdsourcing is a powerful instrument

bull However it is an instrument we do not know thoroughly yet we use ithellip

Therefore

1 Can we help each other to identify ldquoguidelinesrdquo to help us use this instrument in the best way an to its full potentials

2 Is there anything WE can DO to make it even better

Alessandro Bozzon

Alessandro Bozzon

Assistant Professor TU Delft Human-Enhanced Data Management Web Information Systems Social Data Science

Faculty Fellow IBM Benelux Inclusive Enterprise

Investigator AMS Social Sensing Smart Citizens

Web Engineering Web Science

Information Retrieval

User Modelling Crowdsourcing

Human Computation

WWW

ICWE

UMAP

HT

VLDBJ

TWEB

IC

SWJ

JWE

JWS

CSCW

ISWC

Publication Venues

wwwalessandrobozzoncom

Topics of Interest

How can humans and machines better collaborate in solving

(computational) problems

Machines Social

Data

People

Process

A s

ocio

-tech

nic

al syste

m How can human-generated

Web data be transformed

into a source that informs

Web system design

How to enhance Web-

based systems with

automated large-scale

human interpretation

GO

AL

S Correctness

Task Design

Experiment Methodologies

Efficiency Worker Modeling

Getting to Know the Crowd

Sustainability Incentives

Getting to Know the Crowd

Sheelagh Carpendale

Sheelagh Carpendale Canada Research Chair Information Visualization

NSERCAITFSMART Industrial Research Chair Interactive Technologies

InnoVis (Innovations in Visualization)

Interactions Lab

Computer Science University of Calgary

Interactive

Information

Visualization

I have come here to learn

bull While I do a lot of empirical research most is qualitative

bull While I think that crowd sourcing is potentially a fantastic new resource I think successful use of crowd sourcing is non-trivial

bull I hope to learn about how to work with this

Remco Chang

21 Remco Chang ndash Dagstuhl 15

From vision science to data science applying perception to problems in big data

Remco Chang

Assistant Professor Computer Science

Tufts University

22 Remco Chang ndash Dagstuhl 15

Weberrsquos Law

119889119875 = 119896119889119878

119878

Min Chen

Min Chen Professor of Scientific Visualization

University of Oxford (2011 ~ Present) Swansea University (1984 ~ 2011)

Research Interests most aspects of visualization aspects of HCI computer graphics computer vision data mining and information theory

Research Highlight

1 Volume Graphics and Visualization

2 Video Visualization

3 Visual Analytics

4 Perception and Cognition in Visualization

5 Theory of Visualization

1 2

3 4

5

Research Question in the Context of 15481

Short Version How can we use crowdsourcing to capture perceptual and cognitive measures to facilitate mathematical modelling of visualization and interaction

Long Version Information theory is perhaps the most promising mathematical framework that can underpin the fields of visualization and interaction as well as data processing and communication However to obtain its core measures various types of entropy one needs to capture a probability distribution of all states associated with a concept While such a probability distribution can be captured in a machine-centric data processing or communication pipeline it is not trivial to capture in a human-centric process Crowdsourcing offers an opportunity to conduct ldquobigrdquo observation and measurement of perceptual and cognitive activities It is thus a tool for ldquoBig Sciencerdquo

Stephan Diehl

Dagstuhl Seminar 15481 Evaluation in the Crowd Crowdsourcing and Human-Centred Experiments

Stephan Diehl Software Engineering University of Trier Germany

The Java Code Clone

Detection API

Empirical Software Engineering bull [JSS2015 ESEM2015 FSE2014 ICSM 2012

FSE 2011 ESEM 2011 hellip]

Empirical Information Visualization bull [InfoVis 2010 J of Media Psychology 410 hellip]

Analysis and Collaboration SE Tools bull [ASE2014 ASE 2012 ICSE 2011 TSE 2005 ]

Visualizing (Dynamic) Compound Graphs bull [EuroVisSTAR 2014 IV Journal 2012 InfoVis11

SOFTVIS10 EuroVis09 IV09 EuroVis08

CHASE08 AVI08GD04GD02 hellip]

Dagstuhl Seminar 15481 Evaluation in the Crowd Crowdsourcing and Human-Centred Experiments

bull How to reach your (very specific) target audience without

spamming

bull How to control for resolution animation speed color

bull What crowdsourcing platforms exist and what are their

features

Darren Edwards

Darren Edwards Swansea University

Cognitive psychology - categorization

Health and clinical psychology ndash applications of categorization on clinical populations (eg autism)

Crowdsourcing and psychological interventions

Recent validity studies confirm the use of simple psychological tasks such as decision making

Recent attempts have now explored more complex tasks such as psychological interventions (eg mindfulness)

Present studies now explore more complex forms of mindfulness

What are the limitations of using crowdsourced populations online such as when using psychological interventions or categorization

Sebastian Egger

Engineering

Technical high school

bull Telematics

electrotechnical enginieering telecommunications

bull Signal Processing for UWB systems

Sociology

Theory of human societies

Interaction amp collaboration behaviour of groups

Measurement of human behaviour

Currently Human Computer Interaction

Social amp collaborative aspects of technology use

33 26112015

Sebastian Egger

Motivational aspects of crowdworkers

CS as a collaborative work setting

Optimal Setup of CS experiments

Questions (max Nr of questions max duration)

Gamified reliability checks

Scale designs

CS vs Lab whatlsquos the ground truth

New application domains of CS

34 26112015

Interest in Crowdsourcing

Sara Fabrikant

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

humanmdashsystem (visuo-spatial) interface

spatio-temporal analytics

spatialization human navigation

multivariate spatial analysis etc)

interface design

large amp small interactive (map) displays

(ie desktop mobile VR etc)

empirical evaluations

(ie eye tracking EDA EEG etc) [httpwwwunitsevenconz]

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

human sensing observatory set up

controlled laboratory | in-situ (messy) real world setting

semi|automatic data collection of various human behavioral data streams

ie psycho-physiological perceptual cognitive and sensory-motor behaviors (through crowdsourcing )

Brian Fisher

My double research life

Brian Fisher

bull PhD in Experimental Psych

bull Postdoc w Cognitive Science

society founder amp president

bull Psychonomics Fellow

bull VIS-related symposia at Cogsci

amp APS papers on cogsci of

interaction

bull Fuzzy-logicBayes models

bull Postdoc funded by Inst for Robotics

and AI

bull VAST SC VEC VACCINE

bull Co-organized Dagstuhl ldquoInteraction

with Information for Visual Reasoningrdquo

Cogsci-based papers at VIS CHI

BELIV

My Plan

VIS

Me Psychological Science

Ujwal Gadiraju

p = 0055111

Crowd Worker

Behavior

Microtask

Design

Crowdsourcing Paradigm Quality amp Effectiveness

UJWAL GADIRAJU

Incentivization Gamification

Neha Gupta

Getting to know the crowd

Presented at Dagstuhl by

Dr David Martin amp Neha Gupta

Neha Gupta

PhD Student School of Computer Science University of Nottingham UK

Investigating crowdwork through the platform Amazon Mechanical Turk

psxng1nottinghamacuk

Neha_Gi_ji

Research amp Research teams

Research Studies

bull Qualitative study of Turkers on online forums like mturkforum and Turker Nation by Dr David Martin XRCE

bull Ethnographic Studies of Turkers in their respective homes and places of work in India by Neha Gupta UoN

Research teams

Xerox

bull David Martin Jacki OrsquoNeill (Now MSR India) Ben Hanrahan (Now PSU)

psxng1nottinghamacuk

Neha_Gi_ji

Nottingham

bull Andy Crabtree

Tom Rodden

Matthias Hirth

Matthias Hirth 48 48

About Myself

Matthias Hirth

Research Assistant at University of Wuumlrzburg

httpmatthias-hirthcom

matthiashirthinformatikuni-wuerzburgde

Crowdsourcing

Platform analysis

Platform optimization

Human factors

Use cases

Online Social Networks

Collaboration networks

Structural differences of OSNs

Application of OSN relationships

Interactive applicationsGaming

Challenges of real time interactions

Serious gamingGamification

E-Sports broadcastinglive streaming

Quality of Experience

Monitoring techniques

Quantification techniques

(Crowd-bases) test design

Social interactions

Test methodology Incentive design

Use case

Measurement technique

Identification of

qualification

Matthias Hirth 49 49

Improving Crowdsourcing Technology and the Way We Use it

Crowdsourcing technology

How to transfer well known lab procedures into an uncontrolled

crowdsourcing environment

Which crowdsourcing platform features are required to conduct

successful experiments but are not present yet

Which technical developments (can) foster the usage of

crowdsourcing instead of lab experiments

How can technological developments be used for novel incentive

mechanisms

Usage of Crowdsourcing

How can we conduct reproducible experiments in diverse

crowdsourcing environment

How can we raise the awareness of crowdsourcing employers for the

individuals behind the crowd

Tobias Hoszligfeld

The crowd community in the lab

Tobias Hoszligfeld wwwmaswiwiuni-duede

bull Background Tobias Hoszligfeld

ndash 2004-2014 Research group ldquoFuture Internet Applicationsrdquo Chair of Communication Networks Univ of Wuumlrzburg

ndash Since Sep 2014 Chair of Modeling of Adaptive Systems (head) University of Duisburg-Essen

bull Interests Application of crowdsourcing

ndash User studies on Quality of Experience (QoE) for Internet apps eg video QoE in the crowd amp in the lab methodology reliability statisticshellip [2]

ndash Mobile crowdsourcing network and context measurements eg Wifi offloading incentives privacy recommendation mechanisms etc [1]

bull Interests The crowd community

ndash Crowd = Human beings Social network wisdom of the crowd community

ndash UnderstandingUtilizing the crowd community for successful human computation For which jobs can social interaction improve the quality of work Classification

ndash How can we deploy such a system where interaction is triggered

ndash System design Task design interaction mechanisms incentives

22 Nov 2015

[1] httpdropsdagstuhldeopusvolltexte20134354

[2] httpswww3informatikuni-wuerzburgdeqoewikiqualinetcrowd

Challenges Example of Crowdsourced QoE

Tobias Hoszligfeld wwwmaswiwiuni-duede

task design

incentives

Internet

crowd

community

anonymous

remote

subjects

Internet delivery of

test data and user

response

QoE study bull testing methodology bull QoE influence factors

under investigation bull additional QoE factors bull proper subset of items bull hellip

crowdsourcing setting bull short task duration bull instructions amp training bull feedback amp questions bull hidden influence factors bull reliability of users bull hellip

heterogeneous devices

and Internet access

implementation extrinsic vs intrinsic

motivation

Processing of data

Technical aspects bull server implementation bull client application amp

compatibility bull desired test conditions bull context monitoring bull hellip

test items

lab

scientific methodologies

22 Nov 2015

Jason Jacques

Jason Jacques supervised by Per Ola Kristensson

T

Crowdsourcing Code Perception

jtj21camacuk peopledscamacukjtj21

jtjacquesgmailcom jsonjcouk

PhD Working Title

science

Crowdsourcing Interests W

hat

W

ho

Ho

w

Wh

y jtj21camacuk jsonjcouk science

Radu Jianu

Radu Jianu Assistant Professor

Florida International University (PhD Brown University lsquo12)

Interests

Applications of eye-tracking in visualization tracking what people look at rather than where they look

bull Data intensive large scale eye-tracking experiments bull Gaze-contingent visualizations bull Gaze-driven data recommendation and big data curation

Automating user studies in data visualizations (on next page)

Automating user studies

1 Software designs that can (semi)automate a broad range of quantitative evaluations by leveraging

ndash Crowdsourcing ndash Benchmark data and task taxonomies ndash Automated result interpretation

2 Moving user evaluation from the end to within the software design and implementation process as a means to choose between competitive visual designs

Christian Keimel

copy IRT

Where

What

bull Hybrid broadcast broadband technologies

bull Big Data in broadcasting environment and content-creation

bull Audio-visual QoE assessment (with crowdsourcing)

About me

Christian Keimel Slide 60

Lecturer for Digital Broadcast Engineering

Research Engineer for Media Services and Platforms

copy IRT

Crowdsourcing in large scale optimisation

The human component of Big Data

Moving crowdsourcing to lean-back devices

Crowdsourcing from the couch

Crowdsourced experiments emulating the laboratory

Is it time for new methodologies ignoring the lab

Questionshellip

Christian Keimel Slide 61

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 8: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Kathrin Ballweg

Kathrin Ballweg ndash About me

Interests

Computer Science

Cognitive Psychology

User Studies

PhD Focus

Perception and cognition in

network visualization

Visualization for the masses

(journalists and public ndash news readers)

231115 | Dagstuhl Seminar ndash Evaluation in the Crowd Crowdsouring and Human-Centered Experiments | K Ballweg | 9

Kathrin Ballweg

M Sc Computer Science

Technische Universitaumlt

Darmstadt

Design

Evaluation in the Crowd Crowdsourcing and

Human-Centered Experiments

Topics of Interest

Crowdsourcing Platforms vs

The Laboratory

Scientifically Rigorous Methodologies

Crowdsourcing Experiments in Human-

Computer Interaction Visualization and

Applied PerceptionGraphics

231115 | Dagstuhl Seminar ndash Evaluation in the Crowd Crowdsouring and Human-Centered Experiments | K Ballweg | 10

httpwwwtagesnetzwerkde

Rita Borgo

Introduction

Rita Borgo

Visual Computing

Swansea University

Large Data Visualisation HPC

Glyph Based Visualisation

Time Series Analysis

Perception in Visualisation

Visual Computing Group Swansea University

Prof MWJones

D Archamabault R Borgo

R S Laramee B Mora

K W Tam X Xie

My Interest in the Crowdsourcing Phenomena

bull Crowdsourcing is a precious and very appealing resource

bull Crowdsourcing is a powerful instrument

bull However it is an instrument we do not know thoroughly yet we use ithellip

Therefore

1 Can we help each other to identify ldquoguidelinesrdquo to help us use this instrument in the best way an to its full potentials

2 Is there anything WE can DO to make it even better

Alessandro Bozzon

Alessandro Bozzon

Assistant Professor TU Delft Human-Enhanced Data Management Web Information Systems Social Data Science

Faculty Fellow IBM Benelux Inclusive Enterprise

Investigator AMS Social Sensing Smart Citizens

Web Engineering Web Science

Information Retrieval

User Modelling Crowdsourcing

Human Computation

WWW

ICWE

UMAP

HT

VLDBJ

TWEB

IC

SWJ

JWE

JWS

CSCW

ISWC

Publication Venues

wwwalessandrobozzoncom

Topics of Interest

How can humans and machines better collaborate in solving

(computational) problems

Machines Social

Data

People

Process

A s

ocio

-tech

nic

al syste

m How can human-generated

Web data be transformed

into a source that informs

Web system design

How to enhance Web-

based systems with

automated large-scale

human interpretation

GO

AL

S Correctness

Task Design

Experiment Methodologies

Efficiency Worker Modeling

Getting to Know the Crowd

Sustainability Incentives

Getting to Know the Crowd

Sheelagh Carpendale

Sheelagh Carpendale Canada Research Chair Information Visualization

NSERCAITFSMART Industrial Research Chair Interactive Technologies

InnoVis (Innovations in Visualization)

Interactions Lab

Computer Science University of Calgary

Interactive

Information

Visualization

I have come here to learn

bull While I do a lot of empirical research most is qualitative

bull While I think that crowd sourcing is potentially a fantastic new resource I think successful use of crowd sourcing is non-trivial

bull I hope to learn about how to work with this

Remco Chang

21 Remco Chang ndash Dagstuhl 15

From vision science to data science applying perception to problems in big data

Remco Chang

Assistant Professor Computer Science

Tufts University

22 Remco Chang ndash Dagstuhl 15

Weberrsquos Law

119889119875 = 119896119889119878

119878

Min Chen

Min Chen Professor of Scientific Visualization

University of Oxford (2011 ~ Present) Swansea University (1984 ~ 2011)

Research Interests most aspects of visualization aspects of HCI computer graphics computer vision data mining and information theory

Research Highlight

1 Volume Graphics and Visualization

2 Video Visualization

3 Visual Analytics

4 Perception and Cognition in Visualization

5 Theory of Visualization

1 2

3 4

5

Research Question in the Context of 15481

Short Version How can we use crowdsourcing to capture perceptual and cognitive measures to facilitate mathematical modelling of visualization and interaction

Long Version Information theory is perhaps the most promising mathematical framework that can underpin the fields of visualization and interaction as well as data processing and communication However to obtain its core measures various types of entropy one needs to capture a probability distribution of all states associated with a concept While such a probability distribution can be captured in a machine-centric data processing or communication pipeline it is not trivial to capture in a human-centric process Crowdsourcing offers an opportunity to conduct ldquobigrdquo observation and measurement of perceptual and cognitive activities It is thus a tool for ldquoBig Sciencerdquo

Stephan Diehl

Dagstuhl Seminar 15481 Evaluation in the Crowd Crowdsourcing and Human-Centred Experiments

Stephan Diehl Software Engineering University of Trier Germany

The Java Code Clone

Detection API

Empirical Software Engineering bull [JSS2015 ESEM2015 FSE2014 ICSM 2012

FSE 2011 ESEM 2011 hellip]

Empirical Information Visualization bull [InfoVis 2010 J of Media Psychology 410 hellip]

Analysis and Collaboration SE Tools bull [ASE2014 ASE 2012 ICSE 2011 TSE 2005 ]

Visualizing (Dynamic) Compound Graphs bull [EuroVisSTAR 2014 IV Journal 2012 InfoVis11

SOFTVIS10 EuroVis09 IV09 EuroVis08

CHASE08 AVI08GD04GD02 hellip]

Dagstuhl Seminar 15481 Evaluation in the Crowd Crowdsourcing and Human-Centred Experiments

bull How to reach your (very specific) target audience without

spamming

bull How to control for resolution animation speed color

bull What crowdsourcing platforms exist and what are their

features

Darren Edwards

Darren Edwards Swansea University

Cognitive psychology - categorization

Health and clinical psychology ndash applications of categorization on clinical populations (eg autism)

Crowdsourcing and psychological interventions

Recent validity studies confirm the use of simple psychological tasks such as decision making

Recent attempts have now explored more complex tasks such as psychological interventions (eg mindfulness)

Present studies now explore more complex forms of mindfulness

What are the limitations of using crowdsourced populations online such as when using psychological interventions or categorization

Sebastian Egger

Engineering

Technical high school

bull Telematics

electrotechnical enginieering telecommunications

bull Signal Processing for UWB systems

Sociology

Theory of human societies

Interaction amp collaboration behaviour of groups

Measurement of human behaviour

Currently Human Computer Interaction

Social amp collaborative aspects of technology use

33 26112015

Sebastian Egger

Motivational aspects of crowdworkers

CS as a collaborative work setting

Optimal Setup of CS experiments

Questions (max Nr of questions max duration)

Gamified reliability checks

Scale designs

CS vs Lab whatlsquos the ground truth

New application domains of CS

34 26112015

Interest in Crowdsourcing

Sara Fabrikant

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

humanmdashsystem (visuo-spatial) interface

spatio-temporal analytics

spatialization human navigation

multivariate spatial analysis etc)

interface design

large amp small interactive (map) displays

(ie desktop mobile VR etc)

empirical evaluations

(ie eye tracking EDA EEG etc) [httpwwwunitsevenconz]

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

human sensing observatory set up

controlled laboratory | in-situ (messy) real world setting

semi|automatic data collection of various human behavioral data streams

ie psycho-physiological perceptual cognitive and sensory-motor behaviors (through crowdsourcing )

Brian Fisher

My double research life

Brian Fisher

bull PhD in Experimental Psych

bull Postdoc w Cognitive Science

society founder amp president

bull Psychonomics Fellow

bull VIS-related symposia at Cogsci

amp APS papers on cogsci of

interaction

bull Fuzzy-logicBayes models

bull Postdoc funded by Inst for Robotics

and AI

bull VAST SC VEC VACCINE

bull Co-organized Dagstuhl ldquoInteraction

with Information for Visual Reasoningrdquo

Cogsci-based papers at VIS CHI

BELIV

My Plan

VIS

Me Psychological Science

Ujwal Gadiraju

p = 0055111

Crowd Worker

Behavior

Microtask

Design

Crowdsourcing Paradigm Quality amp Effectiveness

UJWAL GADIRAJU

Incentivization Gamification

Neha Gupta

Getting to know the crowd

Presented at Dagstuhl by

Dr David Martin amp Neha Gupta

Neha Gupta

PhD Student School of Computer Science University of Nottingham UK

Investigating crowdwork through the platform Amazon Mechanical Turk

psxng1nottinghamacuk

Neha_Gi_ji

Research amp Research teams

Research Studies

bull Qualitative study of Turkers on online forums like mturkforum and Turker Nation by Dr David Martin XRCE

bull Ethnographic Studies of Turkers in their respective homes and places of work in India by Neha Gupta UoN

Research teams

Xerox

bull David Martin Jacki OrsquoNeill (Now MSR India) Ben Hanrahan (Now PSU)

psxng1nottinghamacuk

Neha_Gi_ji

Nottingham

bull Andy Crabtree

Tom Rodden

Matthias Hirth

Matthias Hirth 48 48

About Myself

Matthias Hirth

Research Assistant at University of Wuumlrzburg

httpmatthias-hirthcom

matthiashirthinformatikuni-wuerzburgde

Crowdsourcing

Platform analysis

Platform optimization

Human factors

Use cases

Online Social Networks

Collaboration networks

Structural differences of OSNs

Application of OSN relationships

Interactive applicationsGaming

Challenges of real time interactions

Serious gamingGamification

E-Sports broadcastinglive streaming

Quality of Experience

Monitoring techniques

Quantification techniques

(Crowd-bases) test design

Social interactions

Test methodology Incentive design

Use case

Measurement technique

Identification of

qualification

Matthias Hirth 49 49

Improving Crowdsourcing Technology and the Way We Use it

Crowdsourcing technology

How to transfer well known lab procedures into an uncontrolled

crowdsourcing environment

Which crowdsourcing platform features are required to conduct

successful experiments but are not present yet

Which technical developments (can) foster the usage of

crowdsourcing instead of lab experiments

How can technological developments be used for novel incentive

mechanisms

Usage of Crowdsourcing

How can we conduct reproducible experiments in diverse

crowdsourcing environment

How can we raise the awareness of crowdsourcing employers for the

individuals behind the crowd

Tobias Hoszligfeld

The crowd community in the lab

Tobias Hoszligfeld wwwmaswiwiuni-duede

bull Background Tobias Hoszligfeld

ndash 2004-2014 Research group ldquoFuture Internet Applicationsrdquo Chair of Communication Networks Univ of Wuumlrzburg

ndash Since Sep 2014 Chair of Modeling of Adaptive Systems (head) University of Duisburg-Essen

bull Interests Application of crowdsourcing

ndash User studies on Quality of Experience (QoE) for Internet apps eg video QoE in the crowd amp in the lab methodology reliability statisticshellip [2]

ndash Mobile crowdsourcing network and context measurements eg Wifi offloading incentives privacy recommendation mechanisms etc [1]

bull Interests The crowd community

ndash Crowd = Human beings Social network wisdom of the crowd community

ndash UnderstandingUtilizing the crowd community for successful human computation For which jobs can social interaction improve the quality of work Classification

ndash How can we deploy such a system where interaction is triggered

ndash System design Task design interaction mechanisms incentives

22 Nov 2015

[1] httpdropsdagstuhldeopusvolltexte20134354

[2] httpswww3informatikuni-wuerzburgdeqoewikiqualinetcrowd

Challenges Example of Crowdsourced QoE

Tobias Hoszligfeld wwwmaswiwiuni-duede

task design

incentives

Internet

crowd

community

anonymous

remote

subjects

Internet delivery of

test data and user

response

QoE study bull testing methodology bull QoE influence factors

under investigation bull additional QoE factors bull proper subset of items bull hellip

crowdsourcing setting bull short task duration bull instructions amp training bull feedback amp questions bull hidden influence factors bull reliability of users bull hellip

heterogeneous devices

and Internet access

implementation extrinsic vs intrinsic

motivation

Processing of data

Technical aspects bull server implementation bull client application amp

compatibility bull desired test conditions bull context monitoring bull hellip

test items

lab

scientific methodologies

22 Nov 2015

Jason Jacques

Jason Jacques supervised by Per Ola Kristensson

T

Crowdsourcing Code Perception

jtj21camacuk peopledscamacukjtj21

jtjacquesgmailcom jsonjcouk

PhD Working Title

science

Crowdsourcing Interests W

hat

W

ho

Ho

w

Wh

y jtj21camacuk jsonjcouk science

Radu Jianu

Radu Jianu Assistant Professor

Florida International University (PhD Brown University lsquo12)

Interests

Applications of eye-tracking in visualization tracking what people look at rather than where they look

bull Data intensive large scale eye-tracking experiments bull Gaze-contingent visualizations bull Gaze-driven data recommendation and big data curation

Automating user studies in data visualizations (on next page)

Automating user studies

1 Software designs that can (semi)automate a broad range of quantitative evaluations by leveraging

ndash Crowdsourcing ndash Benchmark data and task taxonomies ndash Automated result interpretation

2 Moving user evaluation from the end to within the software design and implementation process as a means to choose between competitive visual designs

Christian Keimel

copy IRT

Where

What

bull Hybrid broadcast broadband technologies

bull Big Data in broadcasting environment and content-creation

bull Audio-visual QoE assessment (with crowdsourcing)

About me

Christian Keimel Slide 60

Lecturer for Digital Broadcast Engineering

Research Engineer for Media Services and Platforms

copy IRT

Crowdsourcing in large scale optimisation

The human component of Big Data

Moving crowdsourcing to lean-back devices

Crowdsourcing from the couch

Crowdsourced experiments emulating the laboratory

Is it time for new methodologies ignoring the lab

Questionshellip

Christian Keimel Slide 61

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 9: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Kathrin Ballweg ndash About me

Interests

Computer Science

Cognitive Psychology

User Studies

PhD Focus

Perception and cognition in

network visualization

Visualization for the masses

(journalists and public ndash news readers)

231115 | Dagstuhl Seminar ndash Evaluation in the Crowd Crowdsouring and Human-Centered Experiments | K Ballweg | 9

Kathrin Ballweg

M Sc Computer Science

Technische Universitaumlt

Darmstadt

Design

Evaluation in the Crowd Crowdsourcing and

Human-Centered Experiments

Topics of Interest

Crowdsourcing Platforms vs

The Laboratory

Scientifically Rigorous Methodologies

Crowdsourcing Experiments in Human-

Computer Interaction Visualization and

Applied PerceptionGraphics

231115 | Dagstuhl Seminar ndash Evaluation in the Crowd Crowdsouring and Human-Centered Experiments | K Ballweg | 10

httpwwwtagesnetzwerkde

Rita Borgo

Introduction

Rita Borgo

Visual Computing

Swansea University

Large Data Visualisation HPC

Glyph Based Visualisation

Time Series Analysis

Perception in Visualisation

Visual Computing Group Swansea University

Prof MWJones

D Archamabault R Borgo

R S Laramee B Mora

K W Tam X Xie

My Interest in the Crowdsourcing Phenomena

bull Crowdsourcing is a precious and very appealing resource

bull Crowdsourcing is a powerful instrument

bull However it is an instrument we do not know thoroughly yet we use ithellip

Therefore

1 Can we help each other to identify ldquoguidelinesrdquo to help us use this instrument in the best way an to its full potentials

2 Is there anything WE can DO to make it even better

Alessandro Bozzon

Alessandro Bozzon

Assistant Professor TU Delft Human-Enhanced Data Management Web Information Systems Social Data Science

Faculty Fellow IBM Benelux Inclusive Enterprise

Investigator AMS Social Sensing Smart Citizens

Web Engineering Web Science

Information Retrieval

User Modelling Crowdsourcing

Human Computation

WWW

ICWE

UMAP

HT

VLDBJ

TWEB

IC

SWJ

JWE

JWS

CSCW

ISWC

Publication Venues

wwwalessandrobozzoncom

Topics of Interest

How can humans and machines better collaborate in solving

(computational) problems

Machines Social

Data

People

Process

A s

ocio

-tech

nic

al syste

m How can human-generated

Web data be transformed

into a source that informs

Web system design

How to enhance Web-

based systems with

automated large-scale

human interpretation

GO

AL

S Correctness

Task Design

Experiment Methodologies

Efficiency Worker Modeling

Getting to Know the Crowd

Sustainability Incentives

Getting to Know the Crowd

Sheelagh Carpendale

Sheelagh Carpendale Canada Research Chair Information Visualization

NSERCAITFSMART Industrial Research Chair Interactive Technologies

InnoVis (Innovations in Visualization)

Interactions Lab

Computer Science University of Calgary

Interactive

Information

Visualization

I have come here to learn

bull While I do a lot of empirical research most is qualitative

bull While I think that crowd sourcing is potentially a fantastic new resource I think successful use of crowd sourcing is non-trivial

bull I hope to learn about how to work with this

Remco Chang

21 Remco Chang ndash Dagstuhl 15

From vision science to data science applying perception to problems in big data

Remco Chang

Assistant Professor Computer Science

Tufts University

22 Remco Chang ndash Dagstuhl 15

Weberrsquos Law

119889119875 = 119896119889119878

119878

Min Chen

Min Chen Professor of Scientific Visualization

University of Oxford (2011 ~ Present) Swansea University (1984 ~ 2011)

Research Interests most aspects of visualization aspects of HCI computer graphics computer vision data mining and information theory

Research Highlight

1 Volume Graphics and Visualization

2 Video Visualization

3 Visual Analytics

4 Perception and Cognition in Visualization

5 Theory of Visualization

1 2

3 4

5

Research Question in the Context of 15481

Short Version How can we use crowdsourcing to capture perceptual and cognitive measures to facilitate mathematical modelling of visualization and interaction

Long Version Information theory is perhaps the most promising mathematical framework that can underpin the fields of visualization and interaction as well as data processing and communication However to obtain its core measures various types of entropy one needs to capture a probability distribution of all states associated with a concept While such a probability distribution can be captured in a machine-centric data processing or communication pipeline it is not trivial to capture in a human-centric process Crowdsourcing offers an opportunity to conduct ldquobigrdquo observation and measurement of perceptual and cognitive activities It is thus a tool for ldquoBig Sciencerdquo

Stephan Diehl

Dagstuhl Seminar 15481 Evaluation in the Crowd Crowdsourcing and Human-Centred Experiments

Stephan Diehl Software Engineering University of Trier Germany

The Java Code Clone

Detection API

Empirical Software Engineering bull [JSS2015 ESEM2015 FSE2014 ICSM 2012

FSE 2011 ESEM 2011 hellip]

Empirical Information Visualization bull [InfoVis 2010 J of Media Psychology 410 hellip]

Analysis and Collaboration SE Tools bull [ASE2014 ASE 2012 ICSE 2011 TSE 2005 ]

Visualizing (Dynamic) Compound Graphs bull [EuroVisSTAR 2014 IV Journal 2012 InfoVis11

SOFTVIS10 EuroVis09 IV09 EuroVis08

CHASE08 AVI08GD04GD02 hellip]

Dagstuhl Seminar 15481 Evaluation in the Crowd Crowdsourcing and Human-Centred Experiments

bull How to reach your (very specific) target audience without

spamming

bull How to control for resolution animation speed color

bull What crowdsourcing platforms exist and what are their

features

Darren Edwards

Darren Edwards Swansea University

Cognitive psychology - categorization

Health and clinical psychology ndash applications of categorization on clinical populations (eg autism)

Crowdsourcing and psychological interventions

Recent validity studies confirm the use of simple psychological tasks such as decision making

Recent attempts have now explored more complex tasks such as psychological interventions (eg mindfulness)

Present studies now explore more complex forms of mindfulness

What are the limitations of using crowdsourced populations online such as when using psychological interventions or categorization

Sebastian Egger

Engineering

Technical high school

bull Telematics

electrotechnical enginieering telecommunications

bull Signal Processing for UWB systems

Sociology

Theory of human societies

Interaction amp collaboration behaviour of groups

Measurement of human behaviour

Currently Human Computer Interaction

Social amp collaborative aspects of technology use

33 26112015

Sebastian Egger

Motivational aspects of crowdworkers

CS as a collaborative work setting

Optimal Setup of CS experiments

Questions (max Nr of questions max duration)

Gamified reliability checks

Scale designs

CS vs Lab whatlsquos the ground truth

New application domains of CS

34 26112015

Interest in Crowdsourcing

Sara Fabrikant

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

humanmdashsystem (visuo-spatial) interface

spatio-temporal analytics

spatialization human navigation

multivariate spatial analysis etc)

interface design

large amp small interactive (map) displays

(ie desktop mobile VR etc)

empirical evaluations

(ie eye tracking EDA EEG etc) [httpwwwunitsevenconz]

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

human sensing observatory set up

controlled laboratory | in-situ (messy) real world setting

semi|automatic data collection of various human behavioral data streams

ie psycho-physiological perceptual cognitive and sensory-motor behaviors (through crowdsourcing )

Brian Fisher

My double research life

Brian Fisher

bull PhD in Experimental Psych

bull Postdoc w Cognitive Science

society founder amp president

bull Psychonomics Fellow

bull VIS-related symposia at Cogsci

amp APS papers on cogsci of

interaction

bull Fuzzy-logicBayes models

bull Postdoc funded by Inst for Robotics

and AI

bull VAST SC VEC VACCINE

bull Co-organized Dagstuhl ldquoInteraction

with Information for Visual Reasoningrdquo

Cogsci-based papers at VIS CHI

BELIV

My Plan

VIS

Me Psychological Science

Ujwal Gadiraju

p = 0055111

Crowd Worker

Behavior

Microtask

Design

Crowdsourcing Paradigm Quality amp Effectiveness

UJWAL GADIRAJU

Incentivization Gamification

Neha Gupta

Getting to know the crowd

Presented at Dagstuhl by

Dr David Martin amp Neha Gupta

Neha Gupta

PhD Student School of Computer Science University of Nottingham UK

Investigating crowdwork through the platform Amazon Mechanical Turk

psxng1nottinghamacuk

Neha_Gi_ji

Research amp Research teams

Research Studies

bull Qualitative study of Turkers on online forums like mturkforum and Turker Nation by Dr David Martin XRCE

bull Ethnographic Studies of Turkers in their respective homes and places of work in India by Neha Gupta UoN

Research teams

Xerox

bull David Martin Jacki OrsquoNeill (Now MSR India) Ben Hanrahan (Now PSU)

psxng1nottinghamacuk

Neha_Gi_ji

Nottingham

bull Andy Crabtree

Tom Rodden

Matthias Hirth

Matthias Hirth 48 48

About Myself

Matthias Hirth

Research Assistant at University of Wuumlrzburg

httpmatthias-hirthcom

matthiashirthinformatikuni-wuerzburgde

Crowdsourcing

Platform analysis

Platform optimization

Human factors

Use cases

Online Social Networks

Collaboration networks

Structural differences of OSNs

Application of OSN relationships

Interactive applicationsGaming

Challenges of real time interactions

Serious gamingGamification

E-Sports broadcastinglive streaming

Quality of Experience

Monitoring techniques

Quantification techniques

(Crowd-bases) test design

Social interactions

Test methodology Incentive design

Use case

Measurement technique

Identification of

qualification

Matthias Hirth 49 49

Improving Crowdsourcing Technology and the Way We Use it

Crowdsourcing technology

How to transfer well known lab procedures into an uncontrolled

crowdsourcing environment

Which crowdsourcing platform features are required to conduct

successful experiments but are not present yet

Which technical developments (can) foster the usage of

crowdsourcing instead of lab experiments

How can technological developments be used for novel incentive

mechanisms

Usage of Crowdsourcing

How can we conduct reproducible experiments in diverse

crowdsourcing environment

How can we raise the awareness of crowdsourcing employers for the

individuals behind the crowd

Tobias Hoszligfeld

The crowd community in the lab

Tobias Hoszligfeld wwwmaswiwiuni-duede

bull Background Tobias Hoszligfeld

ndash 2004-2014 Research group ldquoFuture Internet Applicationsrdquo Chair of Communication Networks Univ of Wuumlrzburg

ndash Since Sep 2014 Chair of Modeling of Adaptive Systems (head) University of Duisburg-Essen

bull Interests Application of crowdsourcing

ndash User studies on Quality of Experience (QoE) for Internet apps eg video QoE in the crowd amp in the lab methodology reliability statisticshellip [2]

ndash Mobile crowdsourcing network and context measurements eg Wifi offloading incentives privacy recommendation mechanisms etc [1]

bull Interests The crowd community

ndash Crowd = Human beings Social network wisdom of the crowd community

ndash UnderstandingUtilizing the crowd community for successful human computation For which jobs can social interaction improve the quality of work Classification

ndash How can we deploy such a system where interaction is triggered

ndash System design Task design interaction mechanisms incentives

22 Nov 2015

[1] httpdropsdagstuhldeopusvolltexte20134354

[2] httpswww3informatikuni-wuerzburgdeqoewikiqualinetcrowd

Challenges Example of Crowdsourced QoE

Tobias Hoszligfeld wwwmaswiwiuni-duede

task design

incentives

Internet

crowd

community

anonymous

remote

subjects

Internet delivery of

test data and user

response

QoE study bull testing methodology bull QoE influence factors

under investigation bull additional QoE factors bull proper subset of items bull hellip

crowdsourcing setting bull short task duration bull instructions amp training bull feedback amp questions bull hidden influence factors bull reliability of users bull hellip

heterogeneous devices

and Internet access

implementation extrinsic vs intrinsic

motivation

Processing of data

Technical aspects bull server implementation bull client application amp

compatibility bull desired test conditions bull context monitoring bull hellip

test items

lab

scientific methodologies

22 Nov 2015

Jason Jacques

Jason Jacques supervised by Per Ola Kristensson

T

Crowdsourcing Code Perception

jtj21camacuk peopledscamacukjtj21

jtjacquesgmailcom jsonjcouk

PhD Working Title

science

Crowdsourcing Interests W

hat

W

ho

Ho

w

Wh

y jtj21camacuk jsonjcouk science

Radu Jianu

Radu Jianu Assistant Professor

Florida International University (PhD Brown University lsquo12)

Interests

Applications of eye-tracking in visualization tracking what people look at rather than where they look

bull Data intensive large scale eye-tracking experiments bull Gaze-contingent visualizations bull Gaze-driven data recommendation and big data curation

Automating user studies in data visualizations (on next page)

Automating user studies

1 Software designs that can (semi)automate a broad range of quantitative evaluations by leveraging

ndash Crowdsourcing ndash Benchmark data and task taxonomies ndash Automated result interpretation

2 Moving user evaluation from the end to within the software design and implementation process as a means to choose between competitive visual designs

Christian Keimel

copy IRT

Where

What

bull Hybrid broadcast broadband technologies

bull Big Data in broadcasting environment and content-creation

bull Audio-visual QoE assessment (with crowdsourcing)

About me

Christian Keimel Slide 60

Lecturer for Digital Broadcast Engineering

Research Engineer for Media Services and Platforms

copy IRT

Crowdsourcing in large scale optimisation

The human component of Big Data

Moving crowdsourcing to lean-back devices

Crowdsourcing from the couch

Crowdsourced experiments emulating the laboratory

Is it time for new methodologies ignoring the lab

Questionshellip

Christian Keimel Slide 61

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 10: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Evaluation in the Crowd Crowdsourcing and

Human-Centered Experiments

Topics of Interest

Crowdsourcing Platforms vs

The Laboratory

Scientifically Rigorous Methodologies

Crowdsourcing Experiments in Human-

Computer Interaction Visualization and

Applied PerceptionGraphics

231115 | Dagstuhl Seminar ndash Evaluation in the Crowd Crowdsouring and Human-Centered Experiments | K Ballweg | 10

httpwwwtagesnetzwerkde

Rita Borgo

Introduction

Rita Borgo

Visual Computing

Swansea University

Large Data Visualisation HPC

Glyph Based Visualisation

Time Series Analysis

Perception in Visualisation

Visual Computing Group Swansea University

Prof MWJones

D Archamabault R Borgo

R S Laramee B Mora

K W Tam X Xie

My Interest in the Crowdsourcing Phenomena

bull Crowdsourcing is a precious and very appealing resource

bull Crowdsourcing is a powerful instrument

bull However it is an instrument we do not know thoroughly yet we use ithellip

Therefore

1 Can we help each other to identify ldquoguidelinesrdquo to help us use this instrument in the best way an to its full potentials

2 Is there anything WE can DO to make it even better

Alessandro Bozzon

Alessandro Bozzon

Assistant Professor TU Delft Human-Enhanced Data Management Web Information Systems Social Data Science

Faculty Fellow IBM Benelux Inclusive Enterprise

Investigator AMS Social Sensing Smart Citizens

Web Engineering Web Science

Information Retrieval

User Modelling Crowdsourcing

Human Computation

WWW

ICWE

UMAP

HT

VLDBJ

TWEB

IC

SWJ

JWE

JWS

CSCW

ISWC

Publication Venues

wwwalessandrobozzoncom

Topics of Interest

How can humans and machines better collaborate in solving

(computational) problems

Machines Social

Data

People

Process

A s

ocio

-tech

nic

al syste

m How can human-generated

Web data be transformed

into a source that informs

Web system design

How to enhance Web-

based systems with

automated large-scale

human interpretation

GO

AL

S Correctness

Task Design

Experiment Methodologies

Efficiency Worker Modeling

Getting to Know the Crowd

Sustainability Incentives

Getting to Know the Crowd

Sheelagh Carpendale

Sheelagh Carpendale Canada Research Chair Information Visualization

NSERCAITFSMART Industrial Research Chair Interactive Technologies

InnoVis (Innovations in Visualization)

Interactions Lab

Computer Science University of Calgary

Interactive

Information

Visualization

I have come here to learn

bull While I do a lot of empirical research most is qualitative

bull While I think that crowd sourcing is potentially a fantastic new resource I think successful use of crowd sourcing is non-trivial

bull I hope to learn about how to work with this

Remco Chang

21 Remco Chang ndash Dagstuhl 15

From vision science to data science applying perception to problems in big data

Remco Chang

Assistant Professor Computer Science

Tufts University

22 Remco Chang ndash Dagstuhl 15

Weberrsquos Law

119889119875 = 119896119889119878

119878

Min Chen

Min Chen Professor of Scientific Visualization

University of Oxford (2011 ~ Present) Swansea University (1984 ~ 2011)

Research Interests most aspects of visualization aspects of HCI computer graphics computer vision data mining and information theory

Research Highlight

1 Volume Graphics and Visualization

2 Video Visualization

3 Visual Analytics

4 Perception and Cognition in Visualization

5 Theory of Visualization

1 2

3 4

5

Research Question in the Context of 15481

Short Version How can we use crowdsourcing to capture perceptual and cognitive measures to facilitate mathematical modelling of visualization and interaction

Long Version Information theory is perhaps the most promising mathematical framework that can underpin the fields of visualization and interaction as well as data processing and communication However to obtain its core measures various types of entropy one needs to capture a probability distribution of all states associated with a concept While such a probability distribution can be captured in a machine-centric data processing or communication pipeline it is not trivial to capture in a human-centric process Crowdsourcing offers an opportunity to conduct ldquobigrdquo observation and measurement of perceptual and cognitive activities It is thus a tool for ldquoBig Sciencerdquo

Stephan Diehl

Dagstuhl Seminar 15481 Evaluation in the Crowd Crowdsourcing and Human-Centred Experiments

Stephan Diehl Software Engineering University of Trier Germany

The Java Code Clone

Detection API

Empirical Software Engineering bull [JSS2015 ESEM2015 FSE2014 ICSM 2012

FSE 2011 ESEM 2011 hellip]

Empirical Information Visualization bull [InfoVis 2010 J of Media Psychology 410 hellip]

Analysis and Collaboration SE Tools bull [ASE2014 ASE 2012 ICSE 2011 TSE 2005 ]

Visualizing (Dynamic) Compound Graphs bull [EuroVisSTAR 2014 IV Journal 2012 InfoVis11

SOFTVIS10 EuroVis09 IV09 EuroVis08

CHASE08 AVI08GD04GD02 hellip]

Dagstuhl Seminar 15481 Evaluation in the Crowd Crowdsourcing and Human-Centred Experiments

bull How to reach your (very specific) target audience without

spamming

bull How to control for resolution animation speed color

bull What crowdsourcing platforms exist and what are their

features

Darren Edwards

Darren Edwards Swansea University

Cognitive psychology - categorization

Health and clinical psychology ndash applications of categorization on clinical populations (eg autism)

Crowdsourcing and psychological interventions

Recent validity studies confirm the use of simple psychological tasks such as decision making

Recent attempts have now explored more complex tasks such as psychological interventions (eg mindfulness)

Present studies now explore more complex forms of mindfulness

What are the limitations of using crowdsourced populations online such as when using psychological interventions or categorization

Sebastian Egger

Engineering

Technical high school

bull Telematics

electrotechnical enginieering telecommunications

bull Signal Processing for UWB systems

Sociology

Theory of human societies

Interaction amp collaboration behaviour of groups

Measurement of human behaviour

Currently Human Computer Interaction

Social amp collaborative aspects of technology use

33 26112015

Sebastian Egger

Motivational aspects of crowdworkers

CS as a collaborative work setting

Optimal Setup of CS experiments

Questions (max Nr of questions max duration)

Gamified reliability checks

Scale designs

CS vs Lab whatlsquos the ground truth

New application domains of CS

34 26112015

Interest in Crowdsourcing

Sara Fabrikant

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

humanmdashsystem (visuo-spatial) interface

spatio-temporal analytics

spatialization human navigation

multivariate spatial analysis etc)

interface design

large amp small interactive (map) displays

(ie desktop mobile VR etc)

empirical evaluations

(ie eye tracking EDA EEG etc) [httpwwwunitsevenconz]

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

human sensing observatory set up

controlled laboratory | in-situ (messy) real world setting

semi|automatic data collection of various human behavioral data streams

ie psycho-physiological perceptual cognitive and sensory-motor behaviors (through crowdsourcing )

Brian Fisher

My double research life

Brian Fisher

bull PhD in Experimental Psych

bull Postdoc w Cognitive Science

society founder amp president

bull Psychonomics Fellow

bull VIS-related symposia at Cogsci

amp APS papers on cogsci of

interaction

bull Fuzzy-logicBayes models

bull Postdoc funded by Inst for Robotics

and AI

bull VAST SC VEC VACCINE

bull Co-organized Dagstuhl ldquoInteraction

with Information for Visual Reasoningrdquo

Cogsci-based papers at VIS CHI

BELIV

My Plan

VIS

Me Psychological Science

Ujwal Gadiraju

p = 0055111

Crowd Worker

Behavior

Microtask

Design

Crowdsourcing Paradigm Quality amp Effectiveness

UJWAL GADIRAJU

Incentivization Gamification

Neha Gupta

Getting to know the crowd

Presented at Dagstuhl by

Dr David Martin amp Neha Gupta

Neha Gupta

PhD Student School of Computer Science University of Nottingham UK

Investigating crowdwork through the platform Amazon Mechanical Turk

psxng1nottinghamacuk

Neha_Gi_ji

Research amp Research teams

Research Studies

bull Qualitative study of Turkers on online forums like mturkforum and Turker Nation by Dr David Martin XRCE

bull Ethnographic Studies of Turkers in their respective homes and places of work in India by Neha Gupta UoN

Research teams

Xerox

bull David Martin Jacki OrsquoNeill (Now MSR India) Ben Hanrahan (Now PSU)

psxng1nottinghamacuk

Neha_Gi_ji

Nottingham

bull Andy Crabtree

Tom Rodden

Matthias Hirth

Matthias Hirth 48 48

About Myself

Matthias Hirth

Research Assistant at University of Wuumlrzburg

httpmatthias-hirthcom

matthiashirthinformatikuni-wuerzburgde

Crowdsourcing

Platform analysis

Platform optimization

Human factors

Use cases

Online Social Networks

Collaboration networks

Structural differences of OSNs

Application of OSN relationships

Interactive applicationsGaming

Challenges of real time interactions

Serious gamingGamification

E-Sports broadcastinglive streaming

Quality of Experience

Monitoring techniques

Quantification techniques

(Crowd-bases) test design

Social interactions

Test methodology Incentive design

Use case

Measurement technique

Identification of

qualification

Matthias Hirth 49 49

Improving Crowdsourcing Technology and the Way We Use it

Crowdsourcing technology

How to transfer well known lab procedures into an uncontrolled

crowdsourcing environment

Which crowdsourcing platform features are required to conduct

successful experiments but are not present yet

Which technical developments (can) foster the usage of

crowdsourcing instead of lab experiments

How can technological developments be used for novel incentive

mechanisms

Usage of Crowdsourcing

How can we conduct reproducible experiments in diverse

crowdsourcing environment

How can we raise the awareness of crowdsourcing employers for the

individuals behind the crowd

Tobias Hoszligfeld

The crowd community in the lab

Tobias Hoszligfeld wwwmaswiwiuni-duede

bull Background Tobias Hoszligfeld

ndash 2004-2014 Research group ldquoFuture Internet Applicationsrdquo Chair of Communication Networks Univ of Wuumlrzburg

ndash Since Sep 2014 Chair of Modeling of Adaptive Systems (head) University of Duisburg-Essen

bull Interests Application of crowdsourcing

ndash User studies on Quality of Experience (QoE) for Internet apps eg video QoE in the crowd amp in the lab methodology reliability statisticshellip [2]

ndash Mobile crowdsourcing network and context measurements eg Wifi offloading incentives privacy recommendation mechanisms etc [1]

bull Interests The crowd community

ndash Crowd = Human beings Social network wisdom of the crowd community

ndash UnderstandingUtilizing the crowd community for successful human computation For which jobs can social interaction improve the quality of work Classification

ndash How can we deploy such a system where interaction is triggered

ndash System design Task design interaction mechanisms incentives

22 Nov 2015

[1] httpdropsdagstuhldeopusvolltexte20134354

[2] httpswww3informatikuni-wuerzburgdeqoewikiqualinetcrowd

Challenges Example of Crowdsourced QoE

Tobias Hoszligfeld wwwmaswiwiuni-duede

task design

incentives

Internet

crowd

community

anonymous

remote

subjects

Internet delivery of

test data and user

response

QoE study bull testing methodology bull QoE influence factors

under investigation bull additional QoE factors bull proper subset of items bull hellip

crowdsourcing setting bull short task duration bull instructions amp training bull feedback amp questions bull hidden influence factors bull reliability of users bull hellip

heterogeneous devices

and Internet access

implementation extrinsic vs intrinsic

motivation

Processing of data

Technical aspects bull server implementation bull client application amp

compatibility bull desired test conditions bull context monitoring bull hellip

test items

lab

scientific methodologies

22 Nov 2015

Jason Jacques

Jason Jacques supervised by Per Ola Kristensson

T

Crowdsourcing Code Perception

jtj21camacuk peopledscamacukjtj21

jtjacquesgmailcom jsonjcouk

PhD Working Title

science

Crowdsourcing Interests W

hat

W

ho

Ho

w

Wh

y jtj21camacuk jsonjcouk science

Radu Jianu

Radu Jianu Assistant Professor

Florida International University (PhD Brown University lsquo12)

Interests

Applications of eye-tracking in visualization tracking what people look at rather than where they look

bull Data intensive large scale eye-tracking experiments bull Gaze-contingent visualizations bull Gaze-driven data recommendation and big data curation

Automating user studies in data visualizations (on next page)

Automating user studies

1 Software designs that can (semi)automate a broad range of quantitative evaluations by leveraging

ndash Crowdsourcing ndash Benchmark data and task taxonomies ndash Automated result interpretation

2 Moving user evaluation from the end to within the software design and implementation process as a means to choose between competitive visual designs

Christian Keimel

copy IRT

Where

What

bull Hybrid broadcast broadband technologies

bull Big Data in broadcasting environment and content-creation

bull Audio-visual QoE assessment (with crowdsourcing)

About me

Christian Keimel Slide 60

Lecturer for Digital Broadcast Engineering

Research Engineer for Media Services and Platforms

copy IRT

Crowdsourcing in large scale optimisation

The human component of Big Data

Moving crowdsourcing to lean-back devices

Crowdsourcing from the couch

Crowdsourced experiments emulating the laboratory

Is it time for new methodologies ignoring the lab

Questionshellip

Christian Keimel Slide 61

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 11: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Rita Borgo

Introduction

Rita Borgo

Visual Computing

Swansea University

Large Data Visualisation HPC

Glyph Based Visualisation

Time Series Analysis

Perception in Visualisation

Visual Computing Group Swansea University

Prof MWJones

D Archamabault R Borgo

R S Laramee B Mora

K W Tam X Xie

My Interest in the Crowdsourcing Phenomena

bull Crowdsourcing is a precious and very appealing resource

bull Crowdsourcing is a powerful instrument

bull However it is an instrument we do not know thoroughly yet we use ithellip

Therefore

1 Can we help each other to identify ldquoguidelinesrdquo to help us use this instrument in the best way an to its full potentials

2 Is there anything WE can DO to make it even better

Alessandro Bozzon

Alessandro Bozzon

Assistant Professor TU Delft Human-Enhanced Data Management Web Information Systems Social Data Science

Faculty Fellow IBM Benelux Inclusive Enterprise

Investigator AMS Social Sensing Smart Citizens

Web Engineering Web Science

Information Retrieval

User Modelling Crowdsourcing

Human Computation

WWW

ICWE

UMAP

HT

VLDBJ

TWEB

IC

SWJ

JWE

JWS

CSCW

ISWC

Publication Venues

wwwalessandrobozzoncom

Topics of Interest

How can humans and machines better collaborate in solving

(computational) problems

Machines Social

Data

People

Process

A s

ocio

-tech

nic

al syste

m How can human-generated

Web data be transformed

into a source that informs

Web system design

How to enhance Web-

based systems with

automated large-scale

human interpretation

GO

AL

S Correctness

Task Design

Experiment Methodologies

Efficiency Worker Modeling

Getting to Know the Crowd

Sustainability Incentives

Getting to Know the Crowd

Sheelagh Carpendale

Sheelagh Carpendale Canada Research Chair Information Visualization

NSERCAITFSMART Industrial Research Chair Interactive Technologies

InnoVis (Innovations in Visualization)

Interactions Lab

Computer Science University of Calgary

Interactive

Information

Visualization

I have come here to learn

bull While I do a lot of empirical research most is qualitative

bull While I think that crowd sourcing is potentially a fantastic new resource I think successful use of crowd sourcing is non-trivial

bull I hope to learn about how to work with this

Remco Chang

21 Remco Chang ndash Dagstuhl 15

From vision science to data science applying perception to problems in big data

Remco Chang

Assistant Professor Computer Science

Tufts University

22 Remco Chang ndash Dagstuhl 15

Weberrsquos Law

119889119875 = 119896119889119878

119878

Min Chen

Min Chen Professor of Scientific Visualization

University of Oxford (2011 ~ Present) Swansea University (1984 ~ 2011)

Research Interests most aspects of visualization aspects of HCI computer graphics computer vision data mining and information theory

Research Highlight

1 Volume Graphics and Visualization

2 Video Visualization

3 Visual Analytics

4 Perception and Cognition in Visualization

5 Theory of Visualization

1 2

3 4

5

Research Question in the Context of 15481

Short Version How can we use crowdsourcing to capture perceptual and cognitive measures to facilitate mathematical modelling of visualization and interaction

Long Version Information theory is perhaps the most promising mathematical framework that can underpin the fields of visualization and interaction as well as data processing and communication However to obtain its core measures various types of entropy one needs to capture a probability distribution of all states associated with a concept While such a probability distribution can be captured in a machine-centric data processing or communication pipeline it is not trivial to capture in a human-centric process Crowdsourcing offers an opportunity to conduct ldquobigrdquo observation and measurement of perceptual and cognitive activities It is thus a tool for ldquoBig Sciencerdquo

Stephan Diehl

Dagstuhl Seminar 15481 Evaluation in the Crowd Crowdsourcing and Human-Centred Experiments

Stephan Diehl Software Engineering University of Trier Germany

The Java Code Clone

Detection API

Empirical Software Engineering bull [JSS2015 ESEM2015 FSE2014 ICSM 2012

FSE 2011 ESEM 2011 hellip]

Empirical Information Visualization bull [InfoVis 2010 J of Media Psychology 410 hellip]

Analysis and Collaboration SE Tools bull [ASE2014 ASE 2012 ICSE 2011 TSE 2005 ]

Visualizing (Dynamic) Compound Graphs bull [EuroVisSTAR 2014 IV Journal 2012 InfoVis11

SOFTVIS10 EuroVis09 IV09 EuroVis08

CHASE08 AVI08GD04GD02 hellip]

Dagstuhl Seminar 15481 Evaluation in the Crowd Crowdsourcing and Human-Centred Experiments

bull How to reach your (very specific) target audience without

spamming

bull How to control for resolution animation speed color

bull What crowdsourcing platforms exist and what are their

features

Darren Edwards

Darren Edwards Swansea University

Cognitive psychology - categorization

Health and clinical psychology ndash applications of categorization on clinical populations (eg autism)

Crowdsourcing and psychological interventions

Recent validity studies confirm the use of simple psychological tasks such as decision making

Recent attempts have now explored more complex tasks such as psychological interventions (eg mindfulness)

Present studies now explore more complex forms of mindfulness

What are the limitations of using crowdsourced populations online such as when using psychological interventions or categorization

Sebastian Egger

Engineering

Technical high school

bull Telematics

electrotechnical enginieering telecommunications

bull Signal Processing for UWB systems

Sociology

Theory of human societies

Interaction amp collaboration behaviour of groups

Measurement of human behaviour

Currently Human Computer Interaction

Social amp collaborative aspects of technology use

33 26112015

Sebastian Egger

Motivational aspects of crowdworkers

CS as a collaborative work setting

Optimal Setup of CS experiments

Questions (max Nr of questions max duration)

Gamified reliability checks

Scale designs

CS vs Lab whatlsquos the ground truth

New application domains of CS

34 26112015

Interest in Crowdsourcing

Sara Fabrikant

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

humanmdashsystem (visuo-spatial) interface

spatio-temporal analytics

spatialization human navigation

multivariate spatial analysis etc)

interface design

large amp small interactive (map) displays

(ie desktop mobile VR etc)

empirical evaluations

(ie eye tracking EDA EEG etc) [httpwwwunitsevenconz]

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

human sensing observatory set up

controlled laboratory | in-situ (messy) real world setting

semi|automatic data collection of various human behavioral data streams

ie psycho-physiological perceptual cognitive and sensory-motor behaviors (through crowdsourcing )

Brian Fisher

My double research life

Brian Fisher

bull PhD in Experimental Psych

bull Postdoc w Cognitive Science

society founder amp president

bull Psychonomics Fellow

bull VIS-related symposia at Cogsci

amp APS papers on cogsci of

interaction

bull Fuzzy-logicBayes models

bull Postdoc funded by Inst for Robotics

and AI

bull VAST SC VEC VACCINE

bull Co-organized Dagstuhl ldquoInteraction

with Information for Visual Reasoningrdquo

Cogsci-based papers at VIS CHI

BELIV

My Plan

VIS

Me Psychological Science

Ujwal Gadiraju

p = 0055111

Crowd Worker

Behavior

Microtask

Design

Crowdsourcing Paradigm Quality amp Effectiveness

UJWAL GADIRAJU

Incentivization Gamification

Neha Gupta

Getting to know the crowd

Presented at Dagstuhl by

Dr David Martin amp Neha Gupta

Neha Gupta

PhD Student School of Computer Science University of Nottingham UK

Investigating crowdwork through the platform Amazon Mechanical Turk

psxng1nottinghamacuk

Neha_Gi_ji

Research amp Research teams

Research Studies

bull Qualitative study of Turkers on online forums like mturkforum and Turker Nation by Dr David Martin XRCE

bull Ethnographic Studies of Turkers in their respective homes and places of work in India by Neha Gupta UoN

Research teams

Xerox

bull David Martin Jacki OrsquoNeill (Now MSR India) Ben Hanrahan (Now PSU)

psxng1nottinghamacuk

Neha_Gi_ji

Nottingham

bull Andy Crabtree

Tom Rodden

Matthias Hirth

Matthias Hirth 48 48

About Myself

Matthias Hirth

Research Assistant at University of Wuumlrzburg

httpmatthias-hirthcom

matthiashirthinformatikuni-wuerzburgde

Crowdsourcing

Platform analysis

Platform optimization

Human factors

Use cases

Online Social Networks

Collaboration networks

Structural differences of OSNs

Application of OSN relationships

Interactive applicationsGaming

Challenges of real time interactions

Serious gamingGamification

E-Sports broadcastinglive streaming

Quality of Experience

Monitoring techniques

Quantification techniques

(Crowd-bases) test design

Social interactions

Test methodology Incentive design

Use case

Measurement technique

Identification of

qualification

Matthias Hirth 49 49

Improving Crowdsourcing Technology and the Way We Use it

Crowdsourcing technology

How to transfer well known lab procedures into an uncontrolled

crowdsourcing environment

Which crowdsourcing platform features are required to conduct

successful experiments but are not present yet

Which technical developments (can) foster the usage of

crowdsourcing instead of lab experiments

How can technological developments be used for novel incentive

mechanisms

Usage of Crowdsourcing

How can we conduct reproducible experiments in diverse

crowdsourcing environment

How can we raise the awareness of crowdsourcing employers for the

individuals behind the crowd

Tobias Hoszligfeld

The crowd community in the lab

Tobias Hoszligfeld wwwmaswiwiuni-duede

bull Background Tobias Hoszligfeld

ndash 2004-2014 Research group ldquoFuture Internet Applicationsrdquo Chair of Communication Networks Univ of Wuumlrzburg

ndash Since Sep 2014 Chair of Modeling of Adaptive Systems (head) University of Duisburg-Essen

bull Interests Application of crowdsourcing

ndash User studies on Quality of Experience (QoE) for Internet apps eg video QoE in the crowd amp in the lab methodology reliability statisticshellip [2]

ndash Mobile crowdsourcing network and context measurements eg Wifi offloading incentives privacy recommendation mechanisms etc [1]

bull Interests The crowd community

ndash Crowd = Human beings Social network wisdom of the crowd community

ndash UnderstandingUtilizing the crowd community for successful human computation For which jobs can social interaction improve the quality of work Classification

ndash How can we deploy such a system where interaction is triggered

ndash System design Task design interaction mechanisms incentives

22 Nov 2015

[1] httpdropsdagstuhldeopusvolltexte20134354

[2] httpswww3informatikuni-wuerzburgdeqoewikiqualinetcrowd

Challenges Example of Crowdsourced QoE

Tobias Hoszligfeld wwwmaswiwiuni-duede

task design

incentives

Internet

crowd

community

anonymous

remote

subjects

Internet delivery of

test data and user

response

QoE study bull testing methodology bull QoE influence factors

under investigation bull additional QoE factors bull proper subset of items bull hellip

crowdsourcing setting bull short task duration bull instructions amp training bull feedback amp questions bull hidden influence factors bull reliability of users bull hellip

heterogeneous devices

and Internet access

implementation extrinsic vs intrinsic

motivation

Processing of data

Technical aspects bull server implementation bull client application amp

compatibility bull desired test conditions bull context monitoring bull hellip

test items

lab

scientific methodologies

22 Nov 2015

Jason Jacques

Jason Jacques supervised by Per Ola Kristensson

T

Crowdsourcing Code Perception

jtj21camacuk peopledscamacukjtj21

jtjacquesgmailcom jsonjcouk

PhD Working Title

science

Crowdsourcing Interests W

hat

W

ho

Ho

w

Wh

y jtj21camacuk jsonjcouk science

Radu Jianu

Radu Jianu Assistant Professor

Florida International University (PhD Brown University lsquo12)

Interests

Applications of eye-tracking in visualization tracking what people look at rather than where they look

bull Data intensive large scale eye-tracking experiments bull Gaze-contingent visualizations bull Gaze-driven data recommendation and big data curation

Automating user studies in data visualizations (on next page)

Automating user studies

1 Software designs that can (semi)automate a broad range of quantitative evaluations by leveraging

ndash Crowdsourcing ndash Benchmark data and task taxonomies ndash Automated result interpretation

2 Moving user evaluation from the end to within the software design and implementation process as a means to choose between competitive visual designs

Christian Keimel

copy IRT

Where

What

bull Hybrid broadcast broadband technologies

bull Big Data in broadcasting environment and content-creation

bull Audio-visual QoE assessment (with crowdsourcing)

About me

Christian Keimel Slide 60

Lecturer for Digital Broadcast Engineering

Research Engineer for Media Services and Platforms

copy IRT

Crowdsourcing in large scale optimisation

The human component of Big Data

Moving crowdsourcing to lean-back devices

Crowdsourcing from the couch

Crowdsourced experiments emulating the laboratory

Is it time for new methodologies ignoring the lab

Questionshellip

Christian Keimel Slide 61

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 12: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Introduction

Rita Borgo

Visual Computing

Swansea University

Large Data Visualisation HPC

Glyph Based Visualisation

Time Series Analysis

Perception in Visualisation

Visual Computing Group Swansea University

Prof MWJones

D Archamabault R Borgo

R S Laramee B Mora

K W Tam X Xie

My Interest in the Crowdsourcing Phenomena

bull Crowdsourcing is a precious and very appealing resource

bull Crowdsourcing is a powerful instrument

bull However it is an instrument we do not know thoroughly yet we use ithellip

Therefore

1 Can we help each other to identify ldquoguidelinesrdquo to help us use this instrument in the best way an to its full potentials

2 Is there anything WE can DO to make it even better

Alessandro Bozzon

Alessandro Bozzon

Assistant Professor TU Delft Human-Enhanced Data Management Web Information Systems Social Data Science

Faculty Fellow IBM Benelux Inclusive Enterprise

Investigator AMS Social Sensing Smart Citizens

Web Engineering Web Science

Information Retrieval

User Modelling Crowdsourcing

Human Computation

WWW

ICWE

UMAP

HT

VLDBJ

TWEB

IC

SWJ

JWE

JWS

CSCW

ISWC

Publication Venues

wwwalessandrobozzoncom

Topics of Interest

How can humans and machines better collaborate in solving

(computational) problems

Machines Social

Data

People

Process

A s

ocio

-tech

nic

al syste

m How can human-generated

Web data be transformed

into a source that informs

Web system design

How to enhance Web-

based systems with

automated large-scale

human interpretation

GO

AL

S Correctness

Task Design

Experiment Methodologies

Efficiency Worker Modeling

Getting to Know the Crowd

Sustainability Incentives

Getting to Know the Crowd

Sheelagh Carpendale

Sheelagh Carpendale Canada Research Chair Information Visualization

NSERCAITFSMART Industrial Research Chair Interactive Technologies

InnoVis (Innovations in Visualization)

Interactions Lab

Computer Science University of Calgary

Interactive

Information

Visualization

I have come here to learn

bull While I do a lot of empirical research most is qualitative

bull While I think that crowd sourcing is potentially a fantastic new resource I think successful use of crowd sourcing is non-trivial

bull I hope to learn about how to work with this

Remco Chang

21 Remco Chang ndash Dagstuhl 15

From vision science to data science applying perception to problems in big data

Remco Chang

Assistant Professor Computer Science

Tufts University

22 Remco Chang ndash Dagstuhl 15

Weberrsquos Law

119889119875 = 119896119889119878

119878

Min Chen

Min Chen Professor of Scientific Visualization

University of Oxford (2011 ~ Present) Swansea University (1984 ~ 2011)

Research Interests most aspects of visualization aspects of HCI computer graphics computer vision data mining and information theory

Research Highlight

1 Volume Graphics and Visualization

2 Video Visualization

3 Visual Analytics

4 Perception and Cognition in Visualization

5 Theory of Visualization

1 2

3 4

5

Research Question in the Context of 15481

Short Version How can we use crowdsourcing to capture perceptual and cognitive measures to facilitate mathematical modelling of visualization and interaction

Long Version Information theory is perhaps the most promising mathematical framework that can underpin the fields of visualization and interaction as well as data processing and communication However to obtain its core measures various types of entropy one needs to capture a probability distribution of all states associated with a concept While such a probability distribution can be captured in a machine-centric data processing or communication pipeline it is not trivial to capture in a human-centric process Crowdsourcing offers an opportunity to conduct ldquobigrdquo observation and measurement of perceptual and cognitive activities It is thus a tool for ldquoBig Sciencerdquo

Stephan Diehl

Dagstuhl Seminar 15481 Evaluation in the Crowd Crowdsourcing and Human-Centred Experiments

Stephan Diehl Software Engineering University of Trier Germany

The Java Code Clone

Detection API

Empirical Software Engineering bull [JSS2015 ESEM2015 FSE2014 ICSM 2012

FSE 2011 ESEM 2011 hellip]

Empirical Information Visualization bull [InfoVis 2010 J of Media Psychology 410 hellip]

Analysis and Collaboration SE Tools bull [ASE2014 ASE 2012 ICSE 2011 TSE 2005 ]

Visualizing (Dynamic) Compound Graphs bull [EuroVisSTAR 2014 IV Journal 2012 InfoVis11

SOFTVIS10 EuroVis09 IV09 EuroVis08

CHASE08 AVI08GD04GD02 hellip]

Dagstuhl Seminar 15481 Evaluation in the Crowd Crowdsourcing and Human-Centred Experiments

bull How to reach your (very specific) target audience without

spamming

bull How to control for resolution animation speed color

bull What crowdsourcing platforms exist and what are their

features

Darren Edwards

Darren Edwards Swansea University

Cognitive psychology - categorization

Health and clinical psychology ndash applications of categorization on clinical populations (eg autism)

Crowdsourcing and psychological interventions

Recent validity studies confirm the use of simple psychological tasks such as decision making

Recent attempts have now explored more complex tasks such as psychological interventions (eg mindfulness)

Present studies now explore more complex forms of mindfulness

What are the limitations of using crowdsourced populations online such as when using psychological interventions or categorization

Sebastian Egger

Engineering

Technical high school

bull Telematics

electrotechnical enginieering telecommunications

bull Signal Processing for UWB systems

Sociology

Theory of human societies

Interaction amp collaboration behaviour of groups

Measurement of human behaviour

Currently Human Computer Interaction

Social amp collaborative aspects of technology use

33 26112015

Sebastian Egger

Motivational aspects of crowdworkers

CS as a collaborative work setting

Optimal Setup of CS experiments

Questions (max Nr of questions max duration)

Gamified reliability checks

Scale designs

CS vs Lab whatlsquos the ground truth

New application domains of CS

34 26112015

Interest in Crowdsourcing

Sara Fabrikant

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

humanmdashsystem (visuo-spatial) interface

spatio-temporal analytics

spatialization human navigation

multivariate spatial analysis etc)

interface design

large amp small interactive (map) displays

(ie desktop mobile VR etc)

empirical evaluations

(ie eye tracking EDA EEG etc) [httpwwwunitsevenconz]

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

human sensing observatory set up

controlled laboratory | in-situ (messy) real world setting

semi|automatic data collection of various human behavioral data streams

ie psycho-physiological perceptual cognitive and sensory-motor behaviors (through crowdsourcing )

Brian Fisher

My double research life

Brian Fisher

bull PhD in Experimental Psych

bull Postdoc w Cognitive Science

society founder amp president

bull Psychonomics Fellow

bull VIS-related symposia at Cogsci

amp APS papers on cogsci of

interaction

bull Fuzzy-logicBayes models

bull Postdoc funded by Inst for Robotics

and AI

bull VAST SC VEC VACCINE

bull Co-organized Dagstuhl ldquoInteraction

with Information for Visual Reasoningrdquo

Cogsci-based papers at VIS CHI

BELIV

My Plan

VIS

Me Psychological Science

Ujwal Gadiraju

p = 0055111

Crowd Worker

Behavior

Microtask

Design

Crowdsourcing Paradigm Quality amp Effectiveness

UJWAL GADIRAJU

Incentivization Gamification

Neha Gupta

Getting to know the crowd

Presented at Dagstuhl by

Dr David Martin amp Neha Gupta

Neha Gupta

PhD Student School of Computer Science University of Nottingham UK

Investigating crowdwork through the platform Amazon Mechanical Turk

psxng1nottinghamacuk

Neha_Gi_ji

Research amp Research teams

Research Studies

bull Qualitative study of Turkers on online forums like mturkforum and Turker Nation by Dr David Martin XRCE

bull Ethnographic Studies of Turkers in their respective homes and places of work in India by Neha Gupta UoN

Research teams

Xerox

bull David Martin Jacki OrsquoNeill (Now MSR India) Ben Hanrahan (Now PSU)

psxng1nottinghamacuk

Neha_Gi_ji

Nottingham

bull Andy Crabtree

Tom Rodden

Matthias Hirth

Matthias Hirth 48 48

About Myself

Matthias Hirth

Research Assistant at University of Wuumlrzburg

httpmatthias-hirthcom

matthiashirthinformatikuni-wuerzburgde

Crowdsourcing

Platform analysis

Platform optimization

Human factors

Use cases

Online Social Networks

Collaboration networks

Structural differences of OSNs

Application of OSN relationships

Interactive applicationsGaming

Challenges of real time interactions

Serious gamingGamification

E-Sports broadcastinglive streaming

Quality of Experience

Monitoring techniques

Quantification techniques

(Crowd-bases) test design

Social interactions

Test methodology Incentive design

Use case

Measurement technique

Identification of

qualification

Matthias Hirth 49 49

Improving Crowdsourcing Technology and the Way We Use it

Crowdsourcing technology

How to transfer well known lab procedures into an uncontrolled

crowdsourcing environment

Which crowdsourcing platform features are required to conduct

successful experiments but are not present yet

Which technical developments (can) foster the usage of

crowdsourcing instead of lab experiments

How can technological developments be used for novel incentive

mechanisms

Usage of Crowdsourcing

How can we conduct reproducible experiments in diverse

crowdsourcing environment

How can we raise the awareness of crowdsourcing employers for the

individuals behind the crowd

Tobias Hoszligfeld

The crowd community in the lab

Tobias Hoszligfeld wwwmaswiwiuni-duede

bull Background Tobias Hoszligfeld

ndash 2004-2014 Research group ldquoFuture Internet Applicationsrdquo Chair of Communication Networks Univ of Wuumlrzburg

ndash Since Sep 2014 Chair of Modeling of Adaptive Systems (head) University of Duisburg-Essen

bull Interests Application of crowdsourcing

ndash User studies on Quality of Experience (QoE) for Internet apps eg video QoE in the crowd amp in the lab methodology reliability statisticshellip [2]

ndash Mobile crowdsourcing network and context measurements eg Wifi offloading incentives privacy recommendation mechanisms etc [1]

bull Interests The crowd community

ndash Crowd = Human beings Social network wisdom of the crowd community

ndash UnderstandingUtilizing the crowd community for successful human computation For which jobs can social interaction improve the quality of work Classification

ndash How can we deploy such a system where interaction is triggered

ndash System design Task design interaction mechanisms incentives

22 Nov 2015

[1] httpdropsdagstuhldeopusvolltexte20134354

[2] httpswww3informatikuni-wuerzburgdeqoewikiqualinetcrowd

Challenges Example of Crowdsourced QoE

Tobias Hoszligfeld wwwmaswiwiuni-duede

task design

incentives

Internet

crowd

community

anonymous

remote

subjects

Internet delivery of

test data and user

response

QoE study bull testing methodology bull QoE influence factors

under investigation bull additional QoE factors bull proper subset of items bull hellip

crowdsourcing setting bull short task duration bull instructions amp training bull feedback amp questions bull hidden influence factors bull reliability of users bull hellip

heterogeneous devices

and Internet access

implementation extrinsic vs intrinsic

motivation

Processing of data

Technical aspects bull server implementation bull client application amp

compatibility bull desired test conditions bull context monitoring bull hellip

test items

lab

scientific methodologies

22 Nov 2015

Jason Jacques

Jason Jacques supervised by Per Ola Kristensson

T

Crowdsourcing Code Perception

jtj21camacuk peopledscamacukjtj21

jtjacquesgmailcom jsonjcouk

PhD Working Title

science

Crowdsourcing Interests W

hat

W

ho

Ho

w

Wh

y jtj21camacuk jsonjcouk science

Radu Jianu

Radu Jianu Assistant Professor

Florida International University (PhD Brown University lsquo12)

Interests

Applications of eye-tracking in visualization tracking what people look at rather than where they look

bull Data intensive large scale eye-tracking experiments bull Gaze-contingent visualizations bull Gaze-driven data recommendation and big data curation

Automating user studies in data visualizations (on next page)

Automating user studies

1 Software designs that can (semi)automate a broad range of quantitative evaluations by leveraging

ndash Crowdsourcing ndash Benchmark data and task taxonomies ndash Automated result interpretation

2 Moving user evaluation from the end to within the software design and implementation process as a means to choose between competitive visual designs

Christian Keimel

copy IRT

Where

What

bull Hybrid broadcast broadband technologies

bull Big Data in broadcasting environment and content-creation

bull Audio-visual QoE assessment (with crowdsourcing)

About me

Christian Keimel Slide 60

Lecturer for Digital Broadcast Engineering

Research Engineer for Media Services and Platforms

copy IRT

Crowdsourcing in large scale optimisation

The human component of Big Data

Moving crowdsourcing to lean-back devices

Crowdsourcing from the couch

Crowdsourced experiments emulating the laboratory

Is it time for new methodologies ignoring the lab

Questionshellip

Christian Keimel Slide 61

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 13: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

My Interest in the Crowdsourcing Phenomena

bull Crowdsourcing is a precious and very appealing resource

bull Crowdsourcing is a powerful instrument

bull However it is an instrument we do not know thoroughly yet we use ithellip

Therefore

1 Can we help each other to identify ldquoguidelinesrdquo to help us use this instrument in the best way an to its full potentials

2 Is there anything WE can DO to make it even better

Alessandro Bozzon

Alessandro Bozzon

Assistant Professor TU Delft Human-Enhanced Data Management Web Information Systems Social Data Science

Faculty Fellow IBM Benelux Inclusive Enterprise

Investigator AMS Social Sensing Smart Citizens

Web Engineering Web Science

Information Retrieval

User Modelling Crowdsourcing

Human Computation

WWW

ICWE

UMAP

HT

VLDBJ

TWEB

IC

SWJ

JWE

JWS

CSCW

ISWC

Publication Venues

wwwalessandrobozzoncom

Topics of Interest

How can humans and machines better collaborate in solving

(computational) problems

Machines Social

Data

People

Process

A s

ocio

-tech

nic

al syste

m How can human-generated

Web data be transformed

into a source that informs

Web system design

How to enhance Web-

based systems with

automated large-scale

human interpretation

GO

AL

S Correctness

Task Design

Experiment Methodologies

Efficiency Worker Modeling

Getting to Know the Crowd

Sustainability Incentives

Getting to Know the Crowd

Sheelagh Carpendale

Sheelagh Carpendale Canada Research Chair Information Visualization

NSERCAITFSMART Industrial Research Chair Interactive Technologies

InnoVis (Innovations in Visualization)

Interactions Lab

Computer Science University of Calgary

Interactive

Information

Visualization

I have come here to learn

bull While I do a lot of empirical research most is qualitative

bull While I think that crowd sourcing is potentially a fantastic new resource I think successful use of crowd sourcing is non-trivial

bull I hope to learn about how to work with this

Remco Chang

21 Remco Chang ndash Dagstuhl 15

From vision science to data science applying perception to problems in big data

Remco Chang

Assistant Professor Computer Science

Tufts University

22 Remco Chang ndash Dagstuhl 15

Weberrsquos Law

119889119875 = 119896119889119878

119878

Min Chen

Min Chen Professor of Scientific Visualization

University of Oxford (2011 ~ Present) Swansea University (1984 ~ 2011)

Research Interests most aspects of visualization aspects of HCI computer graphics computer vision data mining and information theory

Research Highlight

1 Volume Graphics and Visualization

2 Video Visualization

3 Visual Analytics

4 Perception and Cognition in Visualization

5 Theory of Visualization

1 2

3 4

5

Research Question in the Context of 15481

Short Version How can we use crowdsourcing to capture perceptual and cognitive measures to facilitate mathematical modelling of visualization and interaction

Long Version Information theory is perhaps the most promising mathematical framework that can underpin the fields of visualization and interaction as well as data processing and communication However to obtain its core measures various types of entropy one needs to capture a probability distribution of all states associated with a concept While such a probability distribution can be captured in a machine-centric data processing or communication pipeline it is not trivial to capture in a human-centric process Crowdsourcing offers an opportunity to conduct ldquobigrdquo observation and measurement of perceptual and cognitive activities It is thus a tool for ldquoBig Sciencerdquo

Stephan Diehl

Dagstuhl Seminar 15481 Evaluation in the Crowd Crowdsourcing and Human-Centred Experiments

Stephan Diehl Software Engineering University of Trier Germany

The Java Code Clone

Detection API

Empirical Software Engineering bull [JSS2015 ESEM2015 FSE2014 ICSM 2012

FSE 2011 ESEM 2011 hellip]

Empirical Information Visualization bull [InfoVis 2010 J of Media Psychology 410 hellip]

Analysis and Collaboration SE Tools bull [ASE2014 ASE 2012 ICSE 2011 TSE 2005 ]

Visualizing (Dynamic) Compound Graphs bull [EuroVisSTAR 2014 IV Journal 2012 InfoVis11

SOFTVIS10 EuroVis09 IV09 EuroVis08

CHASE08 AVI08GD04GD02 hellip]

Dagstuhl Seminar 15481 Evaluation in the Crowd Crowdsourcing and Human-Centred Experiments

bull How to reach your (very specific) target audience without

spamming

bull How to control for resolution animation speed color

bull What crowdsourcing platforms exist and what are their

features

Darren Edwards

Darren Edwards Swansea University

Cognitive psychology - categorization

Health and clinical psychology ndash applications of categorization on clinical populations (eg autism)

Crowdsourcing and psychological interventions

Recent validity studies confirm the use of simple psychological tasks such as decision making

Recent attempts have now explored more complex tasks such as psychological interventions (eg mindfulness)

Present studies now explore more complex forms of mindfulness

What are the limitations of using crowdsourced populations online such as when using psychological interventions or categorization

Sebastian Egger

Engineering

Technical high school

bull Telematics

electrotechnical enginieering telecommunications

bull Signal Processing for UWB systems

Sociology

Theory of human societies

Interaction amp collaboration behaviour of groups

Measurement of human behaviour

Currently Human Computer Interaction

Social amp collaborative aspects of technology use

33 26112015

Sebastian Egger

Motivational aspects of crowdworkers

CS as a collaborative work setting

Optimal Setup of CS experiments

Questions (max Nr of questions max duration)

Gamified reliability checks

Scale designs

CS vs Lab whatlsquos the ground truth

New application domains of CS

34 26112015

Interest in Crowdsourcing

Sara Fabrikant

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

humanmdashsystem (visuo-spatial) interface

spatio-temporal analytics

spatialization human navigation

multivariate spatial analysis etc)

interface design

large amp small interactive (map) displays

(ie desktop mobile VR etc)

empirical evaluations

(ie eye tracking EDA EEG etc) [httpwwwunitsevenconz]

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

human sensing observatory set up

controlled laboratory | in-situ (messy) real world setting

semi|automatic data collection of various human behavioral data streams

ie psycho-physiological perceptual cognitive and sensory-motor behaviors (through crowdsourcing )

Brian Fisher

My double research life

Brian Fisher

bull PhD in Experimental Psych

bull Postdoc w Cognitive Science

society founder amp president

bull Psychonomics Fellow

bull VIS-related symposia at Cogsci

amp APS papers on cogsci of

interaction

bull Fuzzy-logicBayes models

bull Postdoc funded by Inst for Robotics

and AI

bull VAST SC VEC VACCINE

bull Co-organized Dagstuhl ldquoInteraction

with Information for Visual Reasoningrdquo

Cogsci-based papers at VIS CHI

BELIV

My Plan

VIS

Me Psychological Science

Ujwal Gadiraju

p = 0055111

Crowd Worker

Behavior

Microtask

Design

Crowdsourcing Paradigm Quality amp Effectiveness

UJWAL GADIRAJU

Incentivization Gamification

Neha Gupta

Getting to know the crowd

Presented at Dagstuhl by

Dr David Martin amp Neha Gupta

Neha Gupta

PhD Student School of Computer Science University of Nottingham UK

Investigating crowdwork through the platform Amazon Mechanical Turk

psxng1nottinghamacuk

Neha_Gi_ji

Research amp Research teams

Research Studies

bull Qualitative study of Turkers on online forums like mturkforum and Turker Nation by Dr David Martin XRCE

bull Ethnographic Studies of Turkers in their respective homes and places of work in India by Neha Gupta UoN

Research teams

Xerox

bull David Martin Jacki OrsquoNeill (Now MSR India) Ben Hanrahan (Now PSU)

psxng1nottinghamacuk

Neha_Gi_ji

Nottingham

bull Andy Crabtree

Tom Rodden

Matthias Hirth

Matthias Hirth 48 48

About Myself

Matthias Hirth

Research Assistant at University of Wuumlrzburg

httpmatthias-hirthcom

matthiashirthinformatikuni-wuerzburgde

Crowdsourcing

Platform analysis

Platform optimization

Human factors

Use cases

Online Social Networks

Collaboration networks

Structural differences of OSNs

Application of OSN relationships

Interactive applicationsGaming

Challenges of real time interactions

Serious gamingGamification

E-Sports broadcastinglive streaming

Quality of Experience

Monitoring techniques

Quantification techniques

(Crowd-bases) test design

Social interactions

Test methodology Incentive design

Use case

Measurement technique

Identification of

qualification

Matthias Hirth 49 49

Improving Crowdsourcing Technology and the Way We Use it

Crowdsourcing technology

How to transfer well known lab procedures into an uncontrolled

crowdsourcing environment

Which crowdsourcing platform features are required to conduct

successful experiments but are not present yet

Which technical developments (can) foster the usage of

crowdsourcing instead of lab experiments

How can technological developments be used for novel incentive

mechanisms

Usage of Crowdsourcing

How can we conduct reproducible experiments in diverse

crowdsourcing environment

How can we raise the awareness of crowdsourcing employers for the

individuals behind the crowd

Tobias Hoszligfeld

The crowd community in the lab

Tobias Hoszligfeld wwwmaswiwiuni-duede

bull Background Tobias Hoszligfeld

ndash 2004-2014 Research group ldquoFuture Internet Applicationsrdquo Chair of Communication Networks Univ of Wuumlrzburg

ndash Since Sep 2014 Chair of Modeling of Adaptive Systems (head) University of Duisburg-Essen

bull Interests Application of crowdsourcing

ndash User studies on Quality of Experience (QoE) for Internet apps eg video QoE in the crowd amp in the lab methodology reliability statisticshellip [2]

ndash Mobile crowdsourcing network and context measurements eg Wifi offloading incentives privacy recommendation mechanisms etc [1]

bull Interests The crowd community

ndash Crowd = Human beings Social network wisdom of the crowd community

ndash UnderstandingUtilizing the crowd community for successful human computation For which jobs can social interaction improve the quality of work Classification

ndash How can we deploy such a system where interaction is triggered

ndash System design Task design interaction mechanisms incentives

22 Nov 2015

[1] httpdropsdagstuhldeopusvolltexte20134354

[2] httpswww3informatikuni-wuerzburgdeqoewikiqualinetcrowd

Challenges Example of Crowdsourced QoE

Tobias Hoszligfeld wwwmaswiwiuni-duede

task design

incentives

Internet

crowd

community

anonymous

remote

subjects

Internet delivery of

test data and user

response

QoE study bull testing methodology bull QoE influence factors

under investigation bull additional QoE factors bull proper subset of items bull hellip

crowdsourcing setting bull short task duration bull instructions amp training bull feedback amp questions bull hidden influence factors bull reliability of users bull hellip

heterogeneous devices

and Internet access

implementation extrinsic vs intrinsic

motivation

Processing of data

Technical aspects bull server implementation bull client application amp

compatibility bull desired test conditions bull context monitoring bull hellip

test items

lab

scientific methodologies

22 Nov 2015

Jason Jacques

Jason Jacques supervised by Per Ola Kristensson

T

Crowdsourcing Code Perception

jtj21camacuk peopledscamacukjtj21

jtjacquesgmailcom jsonjcouk

PhD Working Title

science

Crowdsourcing Interests W

hat

W

ho

Ho

w

Wh

y jtj21camacuk jsonjcouk science

Radu Jianu

Radu Jianu Assistant Professor

Florida International University (PhD Brown University lsquo12)

Interests

Applications of eye-tracking in visualization tracking what people look at rather than where they look

bull Data intensive large scale eye-tracking experiments bull Gaze-contingent visualizations bull Gaze-driven data recommendation and big data curation

Automating user studies in data visualizations (on next page)

Automating user studies

1 Software designs that can (semi)automate a broad range of quantitative evaluations by leveraging

ndash Crowdsourcing ndash Benchmark data and task taxonomies ndash Automated result interpretation

2 Moving user evaluation from the end to within the software design and implementation process as a means to choose between competitive visual designs

Christian Keimel

copy IRT

Where

What

bull Hybrid broadcast broadband technologies

bull Big Data in broadcasting environment and content-creation

bull Audio-visual QoE assessment (with crowdsourcing)

About me

Christian Keimel Slide 60

Lecturer for Digital Broadcast Engineering

Research Engineer for Media Services and Platforms

copy IRT

Crowdsourcing in large scale optimisation

The human component of Big Data

Moving crowdsourcing to lean-back devices

Crowdsourcing from the couch

Crowdsourced experiments emulating the laboratory

Is it time for new methodologies ignoring the lab

Questionshellip

Christian Keimel Slide 61

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 14: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Alessandro Bozzon

Alessandro Bozzon

Assistant Professor TU Delft Human-Enhanced Data Management Web Information Systems Social Data Science

Faculty Fellow IBM Benelux Inclusive Enterprise

Investigator AMS Social Sensing Smart Citizens

Web Engineering Web Science

Information Retrieval

User Modelling Crowdsourcing

Human Computation

WWW

ICWE

UMAP

HT

VLDBJ

TWEB

IC

SWJ

JWE

JWS

CSCW

ISWC

Publication Venues

wwwalessandrobozzoncom

Topics of Interest

How can humans and machines better collaborate in solving

(computational) problems

Machines Social

Data

People

Process

A s

ocio

-tech

nic

al syste

m How can human-generated

Web data be transformed

into a source that informs

Web system design

How to enhance Web-

based systems with

automated large-scale

human interpretation

GO

AL

S Correctness

Task Design

Experiment Methodologies

Efficiency Worker Modeling

Getting to Know the Crowd

Sustainability Incentives

Getting to Know the Crowd

Sheelagh Carpendale

Sheelagh Carpendale Canada Research Chair Information Visualization

NSERCAITFSMART Industrial Research Chair Interactive Technologies

InnoVis (Innovations in Visualization)

Interactions Lab

Computer Science University of Calgary

Interactive

Information

Visualization

I have come here to learn

bull While I do a lot of empirical research most is qualitative

bull While I think that crowd sourcing is potentially a fantastic new resource I think successful use of crowd sourcing is non-trivial

bull I hope to learn about how to work with this

Remco Chang

21 Remco Chang ndash Dagstuhl 15

From vision science to data science applying perception to problems in big data

Remco Chang

Assistant Professor Computer Science

Tufts University

22 Remco Chang ndash Dagstuhl 15

Weberrsquos Law

119889119875 = 119896119889119878

119878

Min Chen

Min Chen Professor of Scientific Visualization

University of Oxford (2011 ~ Present) Swansea University (1984 ~ 2011)

Research Interests most aspects of visualization aspects of HCI computer graphics computer vision data mining and information theory

Research Highlight

1 Volume Graphics and Visualization

2 Video Visualization

3 Visual Analytics

4 Perception and Cognition in Visualization

5 Theory of Visualization

1 2

3 4

5

Research Question in the Context of 15481

Short Version How can we use crowdsourcing to capture perceptual and cognitive measures to facilitate mathematical modelling of visualization and interaction

Long Version Information theory is perhaps the most promising mathematical framework that can underpin the fields of visualization and interaction as well as data processing and communication However to obtain its core measures various types of entropy one needs to capture a probability distribution of all states associated with a concept While such a probability distribution can be captured in a machine-centric data processing or communication pipeline it is not trivial to capture in a human-centric process Crowdsourcing offers an opportunity to conduct ldquobigrdquo observation and measurement of perceptual and cognitive activities It is thus a tool for ldquoBig Sciencerdquo

Stephan Diehl

Dagstuhl Seminar 15481 Evaluation in the Crowd Crowdsourcing and Human-Centred Experiments

Stephan Diehl Software Engineering University of Trier Germany

The Java Code Clone

Detection API

Empirical Software Engineering bull [JSS2015 ESEM2015 FSE2014 ICSM 2012

FSE 2011 ESEM 2011 hellip]

Empirical Information Visualization bull [InfoVis 2010 J of Media Psychology 410 hellip]

Analysis and Collaboration SE Tools bull [ASE2014 ASE 2012 ICSE 2011 TSE 2005 ]

Visualizing (Dynamic) Compound Graphs bull [EuroVisSTAR 2014 IV Journal 2012 InfoVis11

SOFTVIS10 EuroVis09 IV09 EuroVis08

CHASE08 AVI08GD04GD02 hellip]

Dagstuhl Seminar 15481 Evaluation in the Crowd Crowdsourcing and Human-Centred Experiments

bull How to reach your (very specific) target audience without

spamming

bull How to control for resolution animation speed color

bull What crowdsourcing platforms exist and what are their

features

Darren Edwards

Darren Edwards Swansea University

Cognitive psychology - categorization

Health and clinical psychology ndash applications of categorization on clinical populations (eg autism)

Crowdsourcing and psychological interventions

Recent validity studies confirm the use of simple psychological tasks such as decision making

Recent attempts have now explored more complex tasks such as psychological interventions (eg mindfulness)

Present studies now explore more complex forms of mindfulness

What are the limitations of using crowdsourced populations online such as when using psychological interventions or categorization

Sebastian Egger

Engineering

Technical high school

bull Telematics

electrotechnical enginieering telecommunications

bull Signal Processing for UWB systems

Sociology

Theory of human societies

Interaction amp collaboration behaviour of groups

Measurement of human behaviour

Currently Human Computer Interaction

Social amp collaborative aspects of technology use

33 26112015

Sebastian Egger

Motivational aspects of crowdworkers

CS as a collaborative work setting

Optimal Setup of CS experiments

Questions (max Nr of questions max duration)

Gamified reliability checks

Scale designs

CS vs Lab whatlsquos the ground truth

New application domains of CS

34 26112015

Interest in Crowdsourcing

Sara Fabrikant

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

humanmdashsystem (visuo-spatial) interface

spatio-temporal analytics

spatialization human navigation

multivariate spatial analysis etc)

interface design

large amp small interactive (map) displays

(ie desktop mobile VR etc)

empirical evaluations

(ie eye tracking EDA EEG etc) [httpwwwunitsevenconz]

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

human sensing observatory set up

controlled laboratory | in-situ (messy) real world setting

semi|automatic data collection of various human behavioral data streams

ie psycho-physiological perceptual cognitive and sensory-motor behaviors (through crowdsourcing )

Brian Fisher

My double research life

Brian Fisher

bull PhD in Experimental Psych

bull Postdoc w Cognitive Science

society founder amp president

bull Psychonomics Fellow

bull VIS-related symposia at Cogsci

amp APS papers on cogsci of

interaction

bull Fuzzy-logicBayes models

bull Postdoc funded by Inst for Robotics

and AI

bull VAST SC VEC VACCINE

bull Co-organized Dagstuhl ldquoInteraction

with Information for Visual Reasoningrdquo

Cogsci-based papers at VIS CHI

BELIV

My Plan

VIS

Me Psychological Science

Ujwal Gadiraju

p = 0055111

Crowd Worker

Behavior

Microtask

Design

Crowdsourcing Paradigm Quality amp Effectiveness

UJWAL GADIRAJU

Incentivization Gamification

Neha Gupta

Getting to know the crowd

Presented at Dagstuhl by

Dr David Martin amp Neha Gupta

Neha Gupta

PhD Student School of Computer Science University of Nottingham UK

Investigating crowdwork through the platform Amazon Mechanical Turk

psxng1nottinghamacuk

Neha_Gi_ji

Research amp Research teams

Research Studies

bull Qualitative study of Turkers on online forums like mturkforum and Turker Nation by Dr David Martin XRCE

bull Ethnographic Studies of Turkers in their respective homes and places of work in India by Neha Gupta UoN

Research teams

Xerox

bull David Martin Jacki OrsquoNeill (Now MSR India) Ben Hanrahan (Now PSU)

psxng1nottinghamacuk

Neha_Gi_ji

Nottingham

bull Andy Crabtree

Tom Rodden

Matthias Hirth

Matthias Hirth 48 48

About Myself

Matthias Hirth

Research Assistant at University of Wuumlrzburg

httpmatthias-hirthcom

matthiashirthinformatikuni-wuerzburgde

Crowdsourcing

Platform analysis

Platform optimization

Human factors

Use cases

Online Social Networks

Collaboration networks

Structural differences of OSNs

Application of OSN relationships

Interactive applicationsGaming

Challenges of real time interactions

Serious gamingGamification

E-Sports broadcastinglive streaming

Quality of Experience

Monitoring techniques

Quantification techniques

(Crowd-bases) test design

Social interactions

Test methodology Incentive design

Use case

Measurement technique

Identification of

qualification

Matthias Hirth 49 49

Improving Crowdsourcing Technology and the Way We Use it

Crowdsourcing technology

How to transfer well known lab procedures into an uncontrolled

crowdsourcing environment

Which crowdsourcing platform features are required to conduct

successful experiments but are not present yet

Which technical developments (can) foster the usage of

crowdsourcing instead of lab experiments

How can technological developments be used for novel incentive

mechanisms

Usage of Crowdsourcing

How can we conduct reproducible experiments in diverse

crowdsourcing environment

How can we raise the awareness of crowdsourcing employers for the

individuals behind the crowd

Tobias Hoszligfeld

The crowd community in the lab

Tobias Hoszligfeld wwwmaswiwiuni-duede

bull Background Tobias Hoszligfeld

ndash 2004-2014 Research group ldquoFuture Internet Applicationsrdquo Chair of Communication Networks Univ of Wuumlrzburg

ndash Since Sep 2014 Chair of Modeling of Adaptive Systems (head) University of Duisburg-Essen

bull Interests Application of crowdsourcing

ndash User studies on Quality of Experience (QoE) for Internet apps eg video QoE in the crowd amp in the lab methodology reliability statisticshellip [2]

ndash Mobile crowdsourcing network and context measurements eg Wifi offloading incentives privacy recommendation mechanisms etc [1]

bull Interests The crowd community

ndash Crowd = Human beings Social network wisdom of the crowd community

ndash UnderstandingUtilizing the crowd community for successful human computation For which jobs can social interaction improve the quality of work Classification

ndash How can we deploy such a system where interaction is triggered

ndash System design Task design interaction mechanisms incentives

22 Nov 2015

[1] httpdropsdagstuhldeopusvolltexte20134354

[2] httpswww3informatikuni-wuerzburgdeqoewikiqualinetcrowd

Challenges Example of Crowdsourced QoE

Tobias Hoszligfeld wwwmaswiwiuni-duede

task design

incentives

Internet

crowd

community

anonymous

remote

subjects

Internet delivery of

test data and user

response

QoE study bull testing methodology bull QoE influence factors

under investigation bull additional QoE factors bull proper subset of items bull hellip

crowdsourcing setting bull short task duration bull instructions amp training bull feedback amp questions bull hidden influence factors bull reliability of users bull hellip

heterogeneous devices

and Internet access

implementation extrinsic vs intrinsic

motivation

Processing of data

Technical aspects bull server implementation bull client application amp

compatibility bull desired test conditions bull context monitoring bull hellip

test items

lab

scientific methodologies

22 Nov 2015

Jason Jacques

Jason Jacques supervised by Per Ola Kristensson

T

Crowdsourcing Code Perception

jtj21camacuk peopledscamacukjtj21

jtjacquesgmailcom jsonjcouk

PhD Working Title

science

Crowdsourcing Interests W

hat

W

ho

Ho

w

Wh

y jtj21camacuk jsonjcouk science

Radu Jianu

Radu Jianu Assistant Professor

Florida International University (PhD Brown University lsquo12)

Interests

Applications of eye-tracking in visualization tracking what people look at rather than where they look

bull Data intensive large scale eye-tracking experiments bull Gaze-contingent visualizations bull Gaze-driven data recommendation and big data curation

Automating user studies in data visualizations (on next page)

Automating user studies

1 Software designs that can (semi)automate a broad range of quantitative evaluations by leveraging

ndash Crowdsourcing ndash Benchmark data and task taxonomies ndash Automated result interpretation

2 Moving user evaluation from the end to within the software design and implementation process as a means to choose between competitive visual designs

Christian Keimel

copy IRT

Where

What

bull Hybrid broadcast broadband technologies

bull Big Data in broadcasting environment and content-creation

bull Audio-visual QoE assessment (with crowdsourcing)

About me

Christian Keimel Slide 60

Lecturer for Digital Broadcast Engineering

Research Engineer for Media Services and Platforms

copy IRT

Crowdsourcing in large scale optimisation

The human component of Big Data

Moving crowdsourcing to lean-back devices

Crowdsourcing from the couch

Crowdsourced experiments emulating the laboratory

Is it time for new methodologies ignoring the lab

Questionshellip

Christian Keimel Slide 61

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 15: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Alessandro Bozzon

Assistant Professor TU Delft Human-Enhanced Data Management Web Information Systems Social Data Science

Faculty Fellow IBM Benelux Inclusive Enterprise

Investigator AMS Social Sensing Smart Citizens

Web Engineering Web Science

Information Retrieval

User Modelling Crowdsourcing

Human Computation

WWW

ICWE

UMAP

HT

VLDBJ

TWEB

IC

SWJ

JWE

JWS

CSCW

ISWC

Publication Venues

wwwalessandrobozzoncom

Topics of Interest

How can humans and machines better collaborate in solving

(computational) problems

Machines Social

Data

People

Process

A s

ocio

-tech

nic

al syste

m How can human-generated

Web data be transformed

into a source that informs

Web system design

How to enhance Web-

based systems with

automated large-scale

human interpretation

GO

AL

S Correctness

Task Design

Experiment Methodologies

Efficiency Worker Modeling

Getting to Know the Crowd

Sustainability Incentives

Getting to Know the Crowd

Sheelagh Carpendale

Sheelagh Carpendale Canada Research Chair Information Visualization

NSERCAITFSMART Industrial Research Chair Interactive Technologies

InnoVis (Innovations in Visualization)

Interactions Lab

Computer Science University of Calgary

Interactive

Information

Visualization

I have come here to learn

bull While I do a lot of empirical research most is qualitative

bull While I think that crowd sourcing is potentially a fantastic new resource I think successful use of crowd sourcing is non-trivial

bull I hope to learn about how to work with this

Remco Chang

21 Remco Chang ndash Dagstuhl 15

From vision science to data science applying perception to problems in big data

Remco Chang

Assistant Professor Computer Science

Tufts University

22 Remco Chang ndash Dagstuhl 15

Weberrsquos Law

119889119875 = 119896119889119878

119878

Min Chen

Min Chen Professor of Scientific Visualization

University of Oxford (2011 ~ Present) Swansea University (1984 ~ 2011)

Research Interests most aspects of visualization aspects of HCI computer graphics computer vision data mining and information theory

Research Highlight

1 Volume Graphics and Visualization

2 Video Visualization

3 Visual Analytics

4 Perception and Cognition in Visualization

5 Theory of Visualization

1 2

3 4

5

Research Question in the Context of 15481

Short Version How can we use crowdsourcing to capture perceptual and cognitive measures to facilitate mathematical modelling of visualization and interaction

Long Version Information theory is perhaps the most promising mathematical framework that can underpin the fields of visualization and interaction as well as data processing and communication However to obtain its core measures various types of entropy one needs to capture a probability distribution of all states associated with a concept While such a probability distribution can be captured in a machine-centric data processing or communication pipeline it is not trivial to capture in a human-centric process Crowdsourcing offers an opportunity to conduct ldquobigrdquo observation and measurement of perceptual and cognitive activities It is thus a tool for ldquoBig Sciencerdquo

Stephan Diehl

Dagstuhl Seminar 15481 Evaluation in the Crowd Crowdsourcing and Human-Centred Experiments

Stephan Diehl Software Engineering University of Trier Germany

The Java Code Clone

Detection API

Empirical Software Engineering bull [JSS2015 ESEM2015 FSE2014 ICSM 2012

FSE 2011 ESEM 2011 hellip]

Empirical Information Visualization bull [InfoVis 2010 J of Media Psychology 410 hellip]

Analysis and Collaboration SE Tools bull [ASE2014 ASE 2012 ICSE 2011 TSE 2005 ]

Visualizing (Dynamic) Compound Graphs bull [EuroVisSTAR 2014 IV Journal 2012 InfoVis11

SOFTVIS10 EuroVis09 IV09 EuroVis08

CHASE08 AVI08GD04GD02 hellip]

Dagstuhl Seminar 15481 Evaluation in the Crowd Crowdsourcing and Human-Centred Experiments

bull How to reach your (very specific) target audience without

spamming

bull How to control for resolution animation speed color

bull What crowdsourcing platforms exist and what are their

features

Darren Edwards

Darren Edwards Swansea University

Cognitive psychology - categorization

Health and clinical psychology ndash applications of categorization on clinical populations (eg autism)

Crowdsourcing and psychological interventions

Recent validity studies confirm the use of simple psychological tasks such as decision making

Recent attempts have now explored more complex tasks such as psychological interventions (eg mindfulness)

Present studies now explore more complex forms of mindfulness

What are the limitations of using crowdsourced populations online such as when using psychological interventions or categorization

Sebastian Egger

Engineering

Technical high school

bull Telematics

electrotechnical enginieering telecommunications

bull Signal Processing for UWB systems

Sociology

Theory of human societies

Interaction amp collaboration behaviour of groups

Measurement of human behaviour

Currently Human Computer Interaction

Social amp collaborative aspects of technology use

33 26112015

Sebastian Egger

Motivational aspects of crowdworkers

CS as a collaborative work setting

Optimal Setup of CS experiments

Questions (max Nr of questions max duration)

Gamified reliability checks

Scale designs

CS vs Lab whatlsquos the ground truth

New application domains of CS

34 26112015

Interest in Crowdsourcing

Sara Fabrikant

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

humanmdashsystem (visuo-spatial) interface

spatio-temporal analytics

spatialization human navigation

multivariate spatial analysis etc)

interface design

large amp small interactive (map) displays

(ie desktop mobile VR etc)

empirical evaluations

(ie eye tracking EDA EEG etc) [httpwwwunitsevenconz]

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

human sensing observatory set up

controlled laboratory | in-situ (messy) real world setting

semi|automatic data collection of various human behavioral data streams

ie psycho-physiological perceptual cognitive and sensory-motor behaviors (through crowdsourcing )

Brian Fisher

My double research life

Brian Fisher

bull PhD in Experimental Psych

bull Postdoc w Cognitive Science

society founder amp president

bull Psychonomics Fellow

bull VIS-related symposia at Cogsci

amp APS papers on cogsci of

interaction

bull Fuzzy-logicBayes models

bull Postdoc funded by Inst for Robotics

and AI

bull VAST SC VEC VACCINE

bull Co-organized Dagstuhl ldquoInteraction

with Information for Visual Reasoningrdquo

Cogsci-based papers at VIS CHI

BELIV

My Plan

VIS

Me Psychological Science

Ujwal Gadiraju

p = 0055111

Crowd Worker

Behavior

Microtask

Design

Crowdsourcing Paradigm Quality amp Effectiveness

UJWAL GADIRAJU

Incentivization Gamification

Neha Gupta

Getting to know the crowd

Presented at Dagstuhl by

Dr David Martin amp Neha Gupta

Neha Gupta

PhD Student School of Computer Science University of Nottingham UK

Investigating crowdwork through the platform Amazon Mechanical Turk

psxng1nottinghamacuk

Neha_Gi_ji

Research amp Research teams

Research Studies

bull Qualitative study of Turkers on online forums like mturkforum and Turker Nation by Dr David Martin XRCE

bull Ethnographic Studies of Turkers in their respective homes and places of work in India by Neha Gupta UoN

Research teams

Xerox

bull David Martin Jacki OrsquoNeill (Now MSR India) Ben Hanrahan (Now PSU)

psxng1nottinghamacuk

Neha_Gi_ji

Nottingham

bull Andy Crabtree

Tom Rodden

Matthias Hirth

Matthias Hirth 48 48

About Myself

Matthias Hirth

Research Assistant at University of Wuumlrzburg

httpmatthias-hirthcom

matthiashirthinformatikuni-wuerzburgde

Crowdsourcing

Platform analysis

Platform optimization

Human factors

Use cases

Online Social Networks

Collaboration networks

Structural differences of OSNs

Application of OSN relationships

Interactive applicationsGaming

Challenges of real time interactions

Serious gamingGamification

E-Sports broadcastinglive streaming

Quality of Experience

Monitoring techniques

Quantification techniques

(Crowd-bases) test design

Social interactions

Test methodology Incentive design

Use case

Measurement technique

Identification of

qualification

Matthias Hirth 49 49

Improving Crowdsourcing Technology and the Way We Use it

Crowdsourcing technology

How to transfer well known lab procedures into an uncontrolled

crowdsourcing environment

Which crowdsourcing platform features are required to conduct

successful experiments but are not present yet

Which technical developments (can) foster the usage of

crowdsourcing instead of lab experiments

How can technological developments be used for novel incentive

mechanisms

Usage of Crowdsourcing

How can we conduct reproducible experiments in diverse

crowdsourcing environment

How can we raise the awareness of crowdsourcing employers for the

individuals behind the crowd

Tobias Hoszligfeld

The crowd community in the lab

Tobias Hoszligfeld wwwmaswiwiuni-duede

bull Background Tobias Hoszligfeld

ndash 2004-2014 Research group ldquoFuture Internet Applicationsrdquo Chair of Communication Networks Univ of Wuumlrzburg

ndash Since Sep 2014 Chair of Modeling of Adaptive Systems (head) University of Duisburg-Essen

bull Interests Application of crowdsourcing

ndash User studies on Quality of Experience (QoE) for Internet apps eg video QoE in the crowd amp in the lab methodology reliability statisticshellip [2]

ndash Mobile crowdsourcing network and context measurements eg Wifi offloading incentives privacy recommendation mechanisms etc [1]

bull Interests The crowd community

ndash Crowd = Human beings Social network wisdom of the crowd community

ndash UnderstandingUtilizing the crowd community for successful human computation For which jobs can social interaction improve the quality of work Classification

ndash How can we deploy such a system where interaction is triggered

ndash System design Task design interaction mechanisms incentives

22 Nov 2015

[1] httpdropsdagstuhldeopusvolltexte20134354

[2] httpswww3informatikuni-wuerzburgdeqoewikiqualinetcrowd

Challenges Example of Crowdsourced QoE

Tobias Hoszligfeld wwwmaswiwiuni-duede

task design

incentives

Internet

crowd

community

anonymous

remote

subjects

Internet delivery of

test data and user

response

QoE study bull testing methodology bull QoE influence factors

under investigation bull additional QoE factors bull proper subset of items bull hellip

crowdsourcing setting bull short task duration bull instructions amp training bull feedback amp questions bull hidden influence factors bull reliability of users bull hellip

heterogeneous devices

and Internet access

implementation extrinsic vs intrinsic

motivation

Processing of data

Technical aspects bull server implementation bull client application amp

compatibility bull desired test conditions bull context monitoring bull hellip

test items

lab

scientific methodologies

22 Nov 2015

Jason Jacques

Jason Jacques supervised by Per Ola Kristensson

T

Crowdsourcing Code Perception

jtj21camacuk peopledscamacukjtj21

jtjacquesgmailcom jsonjcouk

PhD Working Title

science

Crowdsourcing Interests W

hat

W

ho

Ho

w

Wh

y jtj21camacuk jsonjcouk science

Radu Jianu

Radu Jianu Assistant Professor

Florida International University (PhD Brown University lsquo12)

Interests

Applications of eye-tracking in visualization tracking what people look at rather than where they look

bull Data intensive large scale eye-tracking experiments bull Gaze-contingent visualizations bull Gaze-driven data recommendation and big data curation

Automating user studies in data visualizations (on next page)

Automating user studies

1 Software designs that can (semi)automate a broad range of quantitative evaluations by leveraging

ndash Crowdsourcing ndash Benchmark data and task taxonomies ndash Automated result interpretation

2 Moving user evaluation from the end to within the software design and implementation process as a means to choose between competitive visual designs

Christian Keimel

copy IRT

Where

What

bull Hybrid broadcast broadband technologies

bull Big Data in broadcasting environment and content-creation

bull Audio-visual QoE assessment (with crowdsourcing)

About me

Christian Keimel Slide 60

Lecturer for Digital Broadcast Engineering

Research Engineer for Media Services and Platforms

copy IRT

Crowdsourcing in large scale optimisation

The human component of Big Data

Moving crowdsourcing to lean-back devices

Crowdsourcing from the couch

Crowdsourced experiments emulating the laboratory

Is it time for new methodologies ignoring the lab

Questionshellip

Christian Keimel Slide 61

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 16: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

How can humans and machines better collaborate in solving

(computational) problems

Machines Social

Data

People

Process

A s

ocio

-tech

nic

al syste

m How can human-generated

Web data be transformed

into a source that informs

Web system design

How to enhance Web-

based systems with

automated large-scale

human interpretation

GO

AL

S Correctness

Task Design

Experiment Methodologies

Efficiency Worker Modeling

Getting to Know the Crowd

Sustainability Incentives

Getting to Know the Crowd

Sheelagh Carpendale

Sheelagh Carpendale Canada Research Chair Information Visualization

NSERCAITFSMART Industrial Research Chair Interactive Technologies

InnoVis (Innovations in Visualization)

Interactions Lab

Computer Science University of Calgary

Interactive

Information

Visualization

I have come here to learn

bull While I do a lot of empirical research most is qualitative

bull While I think that crowd sourcing is potentially a fantastic new resource I think successful use of crowd sourcing is non-trivial

bull I hope to learn about how to work with this

Remco Chang

21 Remco Chang ndash Dagstuhl 15

From vision science to data science applying perception to problems in big data

Remco Chang

Assistant Professor Computer Science

Tufts University

22 Remco Chang ndash Dagstuhl 15

Weberrsquos Law

119889119875 = 119896119889119878

119878

Min Chen

Min Chen Professor of Scientific Visualization

University of Oxford (2011 ~ Present) Swansea University (1984 ~ 2011)

Research Interests most aspects of visualization aspects of HCI computer graphics computer vision data mining and information theory

Research Highlight

1 Volume Graphics and Visualization

2 Video Visualization

3 Visual Analytics

4 Perception and Cognition in Visualization

5 Theory of Visualization

1 2

3 4

5

Research Question in the Context of 15481

Short Version How can we use crowdsourcing to capture perceptual and cognitive measures to facilitate mathematical modelling of visualization and interaction

Long Version Information theory is perhaps the most promising mathematical framework that can underpin the fields of visualization and interaction as well as data processing and communication However to obtain its core measures various types of entropy one needs to capture a probability distribution of all states associated with a concept While such a probability distribution can be captured in a machine-centric data processing or communication pipeline it is not trivial to capture in a human-centric process Crowdsourcing offers an opportunity to conduct ldquobigrdquo observation and measurement of perceptual and cognitive activities It is thus a tool for ldquoBig Sciencerdquo

Stephan Diehl

Dagstuhl Seminar 15481 Evaluation in the Crowd Crowdsourcing and Human-Centred Experiments

Stephan Diehl Software Engineering University of Trier Germany

The Java Code Clone

Detection API

Empirical Software Engineering bull [JSS2015 ESEM2015 FSE2014 ICSM 2012

FSE 2011 ESEM 2011 hellip]

Empirical Information Visualization bull [InfoVis 2010 J of Media Psychology 410 hellip]

Analysis and Collaboration SE Tools bull [ASE2014 ASE 2012 ICSE 2011 TSE 2005 ]

Visualizing (Dynamic) Compound Graphs bull [EuroVisSTAR 2014 IV Journal 2012 InfoVis11

SOFTVIS10 EuroVis09 IV09 EuroVis08

CHASE08 AVI08GD04GD02 hellip]

Dagstuhl Seminar 15481 Evaluation in the Crowd Crowdsourcing and Human-Centred Experiments

bull How to reach your (very specific) target audience without

spamming

bull How to control for resolution animation speed color

bull What crowdsourcing platforms exist and what are their

features

Darren Edwards

Darren Edwards Swansea University

Cognitive psychology - categorization

Health and clinical psychology ndash applications of categorization on clinical populations (eg autism)

Crowdsourcing and psychological interventions

Recent validity studies confirm the use of simple psychological tasks such as decision making

Recent attempts have now explored more complex tasks such as psychological interventions (eg mindfulness)

Present studies now explore more complex forms of mindfulness

What are the limitations of using crowdsourced populations online such as when using psychological interventions or categorization

Sebastian Egger

Engineering

Technical high school

bull Telematics

electrotechnical enginieering telecommunications

bull Signal Processing for UWB systems

Sociology

Theory of human societies

Interaction amp collaboration behaviour of groups

Measurement of human behaviour

Currently Human Computer Interaction

Social amp collaborative aspects of technology use

33 26112015

Sebastian Egger

Motivational aspects of crowdworkers

CS as a collaborative work setting

Optimal Setup of CS experiments

Questions (max Nr of questions max duration)

Gamified reliability checks

Scale designs

CS vs Lab whatlsquos the ground truth

New application domains of CS

34 26112015

Interest in Crowdsourcing

Sara Fabrikant

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

humanmdashsystem (visuo-spatial) interface

spatio-temporal analytics

spatialization human navigation

multivariate spatial analysis etc)

interface design

large amp small interactive (map) displays

(ie desktop mobile VR etc)

empirical evaluations

(ie eye tracking EDA EEG etc) [httpwwwunitsevenconz]

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

human sensing observatory set up

controlled laboratory | in-situ (messy) real world setting

semi|automatic data collection of various human behavioral data streams

ie psycho-physiological perceptual cognitive and sensory-motor behaviors (through crowdsourcing )

Brian Fisher

My double research life

Brian Fisher

bull PhD in Experimental Psych

bull Postdoc w Cognitive Science

society founder amp president

bull Psychonomics Fellow

bull VIS-related symposia at Cogsci

amp APS papers on cogsci of

interaction

bull Fuzzy-logicBayes models

bull Postdoc funded by Inst for Robotics

and AI

bull VAST SC VEC VACCINE

bull Co-organized Dagstuhl ldquoInteraction

with Information for Visual Reasoningrdquo

Cogsci-based papers at VIS CHI

BELIV

My Plan

VIS

Me Psychological Science

Ujwal Gadiraju

p = 0055111

Crowd Worker

Behavior

Microtask

Design

Crowdsourcing Paradigm Quality amp Effectiveness

UJWAL GADIRAJU

Incentivization Gamification

Neha Gupta

Getting to know the crowd

Presented at Dagstuhl by

Dr David Martin amp Neha Gupta

Neha Gupta

PhD Student School of Computer Science University of Nottingham UK

Investigating crowdwork through the platform Amazon Mechanical Turk

psxng1nottinghamacuk

Neha_Gi_ji

Research amp Research teams

Research Studies

bull Qualitative study of Turkers on online forums like mturkforum and Turker Nation by Dr David Martin XRCE

bull Ethnographic Studies of Turkers in their respective homes and places of work in India by Neha Gupta UoN

Research teams

Xerox

bull David Martin Jacki OrsquoNeill (Now MSR India) Ben Hanrahan (Now PSU)

psxng1nottinghamacuk

Neha_Gi_ji

Nottingham

bull Andy Crabtree

Tom Rodden

Matthias Hirth

Matthias Hirth 48 48

About Myself

Matthias Hirth

Research Assistant at University of Wuumlrzburg

httpmatthias-hirthcom

matthiashirthinformatikuni-wuerzburgde

Crowdsourcing

Platform analysis

Platform optimization

Human factors

Use cases

Online Social Networks

Collaboration networks

Structural differences of OSNs

Application of OSN relationships

Interactive applicationsGaming

Challenges of real time interactions

Serious gamingGamification

E-Sports broadcastinglive streaming

Quality of Experience

Monitoring techniques

Quantification techniques

(Crowd-bases) test design

Social interactions

Test methodology Incentive design

Use case

Measurement technique

Identification of

qualification

Matthias Hirth 49 49

Improving Crowdsourcing Technology and the Way We Use it

Crowdsourcing technology

How to transfer well known lab procedures into an uncontrolled

crowdsourcing environment

Which crowdsourcing platform features are required to conduct

successful experiments but are not present yet

Which technical developments (can) foster the usage of

crowdsourcing instead of lab experiments

How can technological developments be used for novel incentive

mechanisms

Usage of Crowdsourcing

How can we conduct reproducible experiments in diverse

crowdsourcing environment

How can we raise the awareness of crowdsourcing employers for the

individuals behind the crowd

Tobias Hoszligfeld

The crowd community in the lab

Tobias Hoszligfeld wwwmaswiwiuni-duede

bull Background Tobias Hoszligfeld

ndash 2004-2014 Research group ldquoFuture Internet Applicationsrdquo Chair of Communication Networks Univ of Wuumlrzburg

ndash Since Sep 2014 Chair of Modeling of Adaptive Systems (head) University of Duisburg-Essen

bull Interests Application of crowdsourcing

ndash User studies on Quality of Experience (QoE) for Internet apps eg video QoE in the crowd amp in the lab methodology reliability statisticshellip [2]

ndash Mobile crowdsourcing network and context measurements eg Wifi offloading incentives privacy recommendation mechanisms etc [1]

bull Interests The crowd community

ndash Crowd = Human beings Social network wisdom of the crowd community

ndash UnderstandingUtilizing the crowd community for successful human computation For which jobs can social interaction improve the quality of work Classification

ndash How can we deploy such a system where interaction is triggered

ndash System design Task design interaction mechanisms incentives

22 Nov 2015

[1] httpdropsdagstuhldeopusvolltexte20134354

[2] httpswww3informatikuni-wuerzburgdeqoewikiqualinetcrowd

Challenges Example of Crowdsourced QoE

Tobias Hoszligfeld wwwmaswiwiuni-duede

task design

incentives

Internet

crowd

community

anonymous

remote

subjects

Internet delivery of

test data and user

response

QoE study bull testing methodology bull QoE influence factors

under investigation bull additional QoE factors bull proper subset of items bull hellip

crowdsourcing setting bull short task duration bull instructions amp training bull feedback amp questions bull hidden influence factors bull reliability of users bull hellip

heterogeneous devices

and Internet access

implementation extrinsic vs intrinsic

motivation

Processing of data

Technical aspects bull server implementation bull client application amp

compatibility bull desired test conditions bull context monitoring bull hellip

test items

lab

scientific methodologies

22 Nov 2015

Jason Jacques

Jason Jacques supervised by Per Ola Kristensson

T

Crowdsourcing Code Perception

jtj21camacuk peopledscamacukjtj21

jtjacquesgmailcom jsonjcouk

PhD Working Title

science

Crowdsourcing Interests W

hat

W

ho

Ho

w

Wh

y jtj21camacuk jsonjcouk science

Radu Jianu

Radu Jianu Assistant Professor

Florida International University (PhD Brown University lsquo12)

Interests

Applications of eye-tracking in visualization tracking what people look at rather than where they look

bull Data intensive large scale eye-tracking experiments bull Gaze-contingent visualizations bull Gaze-driven data recommendation and big data curation

Automating user studies in data visualizations (on next page)

Automating user studies

1 Software designs that can (semi)automate a broad range of quantitative evaluations by leveraging

ndash Crowdsourcing ndash Benchmark data and task taxonomies ndash Automated result interpretation

2 Moving user evaluation from the end to within the software design and implementation process as a means to choose between competitive visual designs

Christian Keimel

copy IRT

Where

What

bull Hybrid broadcast broadband technologies

bull Big Data in broadcasting environment and content-creation

bull Audio-visual QoE assessment (with crowdsourcing)

About me

Christian Keimel Slide 60

Lecturer for Digital Broadcast Engineering

Research Engineer for Media Services and Platforms

copy IRT

Crowdsourcing in large scale optimisation

The human component of Big Data

Moving crowdsourcing to lean-back devices

Crowdsourcing from the couch

Crowdsourced experiments emulating the laboratory

Is it time for new methodologies ignoring the lab

Questionshellip

Christian Keimel Slide 61

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 17: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Sheelagh Carpendale

Sheelagh Carpendale Canada Research Chair Information Visualization

NSERCAITFSMART Industrial Research Chair Interactive Technologies

InnoVis (Innovations in Visualization)

Interactions Lab

Computer Science University of Calgary

Interactive

Information

Visualization

I have come here to learn

bull While I do a lot of empirical research most is qualitative

bull While I think that crowd sourcing is potentially a fantastic new resource I think successful use of crowd sourcing is non-trivial

bull I hope to learn about how to work with this

Remco Chang

21 Remco Chang ndash Dagstuhl 15

From vision science to data science applying perception to problems in big data

Remco Chang

Assistant Professor Computer Science

Tufts University

22 Remco Chang ndash Dagstuhl 15

Weberrsquos Law

119889119875 = 119896119889119878

119878

Min Chen

Min Chen Professor of Scientific Visualization

University of Oxford (2011 ~ Present) Swansea University (1984 ~ 2011)

Research Interests most aspects of visualization aspects of HCI computer graphics computer vision data mining and information theory

Research Highlight

1 Volume Graphics and Visualization

2 Video Visualization

3 Visual Analytics

4 Perception and Cognition in Visualization

5 Theory of Visualization

1 2

3 4

5

Research Question in the Context of 15481

Short Version How can we use crowdsourcing to capture perceptual and cognitive measures to facilitate mathematical modelling of visualization and interaction

Long Version Information theory is perhaps the most promising mathematical framework that can underpin the fields of visualization and interaction as well as data processing and communication However to obtain its core measures various types of entropy one needs to capture a probability distribution of all states associated with a concept While such a probability distribution can be captured in a machine-centric data processing or communication pipeline it is not trivial to capture in a human-centric process Crowdsourcing offers an opportunity to conduct ldquobigrdquo observation and measurement of perceptual and cognitive activities It is thus a tool for ldquoBig Sciencerdquo

Stephan Diehl

Dagstuhl Seminar 15481 Evaluation in the Crowd Crowdsourcing and Human-Centred Experiments

Stephan Diehl Software Engineering University of Trier Germany

The Java Code Clone

Detection API

Empirical Software Engineering bull [JSS2015 ESEM2015 FSE2014 ICSM 2012

FSE 2011 ESEM 2011 hellip]

Empirical Information Visualization bull [InfoVis 2010 J of Media Psychology 410 hellip]

Analysis and Collaboration SE Tools bull [ASE2014 ASE 2012 ICSE 2011 TSE 2005 ]

Visualizing (Dynamic) Compound Graphs bull [EuroVisSTAR 2014 IV Journal 2012 InfoVis11

SOFTVIS10 EuroVis09 IV09 EuroVis08

CHASE08 AVI08GD04GD02 hellip]

Dagstuhl Seminar 15481 Evaluation in the Crowd Crowdsourcing and Human-Centred Experiments

bull How to reach your (very specific) target audience without

spamming

bull How to control for resolution animation speed color

bull What crowdsourcing platforms exist and what are their

features

Darren Edwards

Darren Edwards Swansea University

Cognitive psychology - categorization

Health and clinical psychology ndash applications of categorization on clinical populations (eg autism)

Crowdsourcing and psychological interventions

Recent validity studies confirm the use of simple psychological tasks such as decision making

Recent attempts have now explored more complex tasks such as psychological interventions (eg mindfulness)

Present studies now explore more complex forms of mindfulness

What are the limitations of using crowdsourced populations online such as when using psychological interventions or categorization

Sebastian Egger

Engineering

Technical high school

bull Telematics

electrotechnical enginieering telecommunications

bull Signal Processing for UWB systems

Sociology

Theory of human societies

Interaction amp collaboration behaviour of groups

Measurement of human behaviour

Currently Human Computer Interaction

Social amp collaborative aspects of technology use

33 26112015

Sebastian Egger

Motivational aspects of crowdworkers

CS as a collaborative work setting

Optimal Setup of CS experiments

Questions (max Nr of questions max duration)

Gamified reliability checks

Scale designs

CS vs Lab whatlsquos the ground truth

New application domains of CS

34 26112015

Interest in Crowdsourcing

Sara Fabrikant

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

humanmdashsystem (visuo-spatial) interface

spatio-temporal analytics

spatialization human navigation

multivariate spatial analysis etc)

interface design

large amp small interactive (map) displays

(ie desktop mobile VR etc)

empirical evaluations

(ie eye tracking EDA EEG etc) [httpwwwunitsevenconz]

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

human sensing observatory set up

controlled laboratory | in-situ (messy) real world setting

semi|automatic data collection of various human behavioral data streams

ie psycho-physiological perceptual cognitive and sensory-motor behaviors (through crowdsourcing )

Brian Fisher

My double research life

Brian Fisher

bull PhD in Experimental Psych

bull Postdoc w Cognitive Science

society founder amp president

bull Psychonomics Fellow

bull VIS-related symposia at Cogsci

amp APS papers on cogsci of

interaction

bull Fuzzy-logicBayes models

bull Postdoc funded by Inst for Robotics

and AI

bull VAST SC VEC VACCINE

bull Co-organized Dagstuhl ldquoInteraction

with Information for Visual Reasoningrdquo

Cogsci-based papers at VIS CHI

BELIV

My Plan

VIS

Me Psychological Science

Ujwal Gadiraju

p = 0055111

Crowd Worker

Behavior

Microtask

Design

Crowdsourcing Paradigm Quality amp Effectiveness

UJWAL GADIRAJU

Incentivization Gamification

Neha Gupta

Getting to know the crowd

Presented at Dagstuhl by

Dr David Martin amp Neha Gupta

Neha Gupta

PhD Student School of Computer Science University of Nottingham UK

Investigating crowdwork through the platform Amazon Mechanical Turk

psxng1nottinghamacuk

Neha_Gi_ji

Research amp Research teams

Research Studies

bull Qualitative study of Turkers on online forums like mturkforum and Turker Nation by Dr David Martin XRCE

bull Ethnographic Studies of Turkers in their respective homes and places of work in India by Neha Gupta UoN

Research teams

Xerox

bull David Martin Jacki OrsquoNeill (Now MSR India) Ben Hanrahan (Now PSU)

psxng1nottinghamacuk

Neha_Gi_ji

Nottingham

bull Andy Crabtree

Tom Rodden

Matthias Hirth

Matthias Hirth 48 48

About Myself

Matthias Hirth

Research Assistant at University of Wuumlrzburg

httpmatthias-hirthcom

matthiashirthinformatikuni-wuerzburgde

Crowdsourcing

Platform analysis

Platform optimization

Human factors

Use cases

Online Social Networks

Collaboration networks

Structural differences of OSNs

Application of OSN relationships

Interactive applicationsGaming

Challenges of real time interactions

Serious gamingGamification

E-Sports broadcastinglive streaming

Quality of Experience

Monitoring techniques

Quantification techniques

(Crowd-bases) test design

Social interactions

Test methodology Incentive design

Use case

Measurement technique

Identification of

qualification

Matthias Hirth 49 49

Improving Crowdsourcing Technology and the Way We Use it

Crowdsourcing technology

How to transfer well known lab procedures into an uncontrolled

crowdsourcing environment

Which crowdsourcing platform features are required to conduct

successful experiments but are not present yet

Which technical developments (can) foster the usage of

crowdsourcing instead of lab experiments

How can technological developments be used for novel incentive

mechanisms

Usage of Crowdsourcing

How can we conduct reproducible experiments in diverse

crowdsourcing environment

How can we raise the awareness of crowdsourcing employers for the

individuals behind the crowd

Tobias Hoszligfeld

The crowd community in the lab

Tobias Hoszligfeld wwwmaswiwiuni-duede

bull Background Tobias Hoszligfeld

ndash 2004-2014 Research group ldquoFuture Internet Applicationsrdquo Chair of Communication Networks Univ of Wuumlrzburg

ndash Since Sep 2014 Chair of Modeling of Adaptive Systems (head) University of Duisburg-Essen

bull Interests Application of crowdsourcing

ndash User studies on Quality of Experience (QoE) for Internet apps eg video QoE in the crowd amp in the lab methodology reliability statisticshellip [2]

ndash Mobile crowdsourcing network and context measurements eg Wifi offloading incentives privacy recommendation mechanisms etc [1]

bull Interests The crowd community

ndash Crowd = Human beings Social network wisdom of the crowd community

ndash UnderstandingUtilizing the crowd community for successful human computation For which jobs can social interaction improve the quality of work Classification

ndash How can we deploy such a system where interaction is triggered

ndash System design Task design interaction mechanisms incentives

22 Nov 2015

[1] httpdropsdagstuhldeopusvolltexte20134354

[2] httpswww3informatikuni-wuerzburgdeqoewikiqualinetcrowd

Challenges Example of Crowdsourced QoE

Tobias Hoszligfeld wwwmaswiwiuni-duede

task design

incentives

Internet

crowd

community

anonymous

remote

subjects

Internet delivery of

test data and user

response

QoE study bull testing methodology bull QoE influence factors

under investigation bull additional QoE factors bull proper subset of items bull hellip

crowdsourcing setting bull short task duration bull instructions amp training bull feedback amp questions bull hidden influence factors bull reliability of users bull hellip

heterogeneous devices

and Internet access

implementation extrinsic vs intrinsic

motivation

Processing of data

Technical aspects bull server implementation bull client application amp

compatibility bull desired test conditions bull context monitoring bull hellip

test items

lab

scientific methodologies

22 Nov 2015

Jason Jacques

Jason Jacques supervised by Per Ola Kristensson

T

Crowdsourcing Code Perception

jtj21camacuk peopledscamacukjtj21

jtjacquesgmailcom jsonjcouk

PhD Working Title

science

Crowdsourcing Interests W

hat

W

ho

Ho

w

Wh

y jtj21camacuk jsonjcouk science

Radu Jianu

Radu Jianu Assistant Professor

Florida International University (PhD Brown University lsquo12)

Interests

Applications of eye-tracking in visualization tracking what people look at rather than where they look

bull Data intensive large scale eye-tracking experiments bull Gaze-contingent visualizations bull Gaze-driven data recommendation and big data curation

Automating user studies in data visualizations (on next page)

Automating user studies

1 Software designs that can (semi)automate a broad range of quantitative evaluations by leveraging

ndash Crowdsourcing ndash Benchmark data and task taxonomies ndash Automated result interpretation

2 Moving user evaluation from the end to within the software design and implementation process as a means to choose between competitive visual designs

Christian Keimel

copy IRT

Where

What

bull Hybrid broadcast broadband technologies

bull Big Data in broadcasting environment and content-creation

bull Audio-visual QoE assessment (with crowdsourcing)

About me

Christian Keimel Slide 60

Lecturer for Digital Broadcast Engineering

Research Engineer for Media Services and Platforms

copy IRT

Crowdsourcing in large scale optimisation

The human component of Big Data

Moving crowdsourcing to lean-back devices

Crowdsourcing from the couch

Crowdsourced experiments emulating the laboratory

Is it time for new methodologies ignoring the lab

Questionshellip

Christian Keimel Slide 61

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 18: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Sheelagh Carpendale Canada Research Chair Information Visualization

NSERCAITFSMART Industrial Research Chair Interactive Technologies

InnoVis (Innovations in Visualization)

Interactions Lab

Computer Science University of Calgary

Interactive

Information

Visualization

I have come here to learn

bull While I do a lot of empirical research most is qualitative

bull While I think that crowd sourcing is potentially a fantastic new resource I think successful use of crowd sourcing is non-trivial

bull I hope to learn about how to work with this

Remco Chang

21 Remco Chang ndash Dagstuhl 15

From vision science to data science applying perception to problems in big data

Remco Chang

Assistant Professor Computer Science

Tufts University

22 Remco Chang ndash Dagstuhl 15

Weberrsquos Law

119889119875 = 119896119889119878

119878

Min Chen

Min Chen Professor of Scientific Visualization

University of Oxford (2011 ~ Present) Swansea University (1984 ~ 2011)

Research Interests most aspects of visualization aspects of HCI computer graphics computer vision data mining and information theory

Research Highlight

1 Volume Graphics and Visualization

2 Video Visualization

3 Visual Analytics

4 Perception and Cognition in Visualization

5 Theory of Visualization

1 2

3 4

5

Research Question in the Context of 15481

Short Version How can we use crowdsourcing to capture perceptual and cognitive measures to facilitate mathematical modelling of visualization and interaction

Long Version Information theory is perhaps the most promising mathematical framework that can underpin the fields of visualization and interaction as well as data processing and communication However to obtain its core measures various types of entropy one needs to capture a probability distribution of all states associated with a concept While such a probability distribution can be captured in a machine-centric data processing or communication pipeline it is not trivial to capture in a human-centric process Crowdsourcing offers an opportunity to conduct ldquobigrdquo observation and measurement of perceptual and cognitive activities It is thus a tool for ldquoBig Sciencerdquo

Stephan Diehl

Dagstuhl Seminar 15481 Evaluation in the Crowd Crowdsourcing and Human-Centred Experiments

Stephan Diehl Software Engineering University of Trier Germany

The Java Code Clone

Detection API

Empirical Software Engineering bull [JSS2015 ESEM2015 FSE2014 ICSM 2012

FSE 2011 ESEM 2011 hellip]

Empirical Information Visualization bull [InfoVis 2010 J of Media Psychology 410 hellip]

Analysis and Collaboration SE Tools bull [ASE2014 ASE 2012 ICSE 2011 TSE 2005 ]

Visualizing (Dynamic) Compound Graphs bull [EuroVisSTAR 2014 IV Journal 2012 InfoVis11

SOFTVIS10 EuroVis09 IV09 EuroVis08

CHASE08 AVI08GD04GD02 hellip]

Dagstuhl Seminar 15481 Evaluation in the Crowd Crowdsourcing and Human-Centred Experiments

bull How to reach your (very specific) target audience without

spamming

bull How to control for resolution animation speed color

bull What crowdsourcing platforms exist and what are their

features

Darren Edwards

Darren Edwards Swansea University

Cognitive psychology - categorization

Health and clinical psychology ndash applications of categorization on clinical populations (eg autism)

Crowdsourcing and psychological interventions

Recent validity studies confirm the use of simple psychological tasks such as decision making

Recent attempts have now explored more complex tasks such as psychological interventions (eg mindfulness)

Present studies now explore more complex forms of mindfulness

What are the limitations of using crowdsourced populations online such as when using psychological interventions or categorization

Sebastian Egger

Engineering

Technical high school

bull Telematics

electrotechnical enginieering telecommunications

bull Signal Processing for UWB systems

Sociology

Theory of human societies

Interaction amp collaboration behaviour of groups

Measurement of human behaviour

Currently Human Computer Interaction

Social amp collaborative aspects of technology use

33 26112015

Sebastian Egger

Motivational aspects of crowdworkers

CS as a collaborative work setting

Optimal Setup of CS experiments

Questions (max Nr of questions max duration)

Gamified reliability checks

Scale designs

CS vs Lab whatlsquos the ground truth

New application domains of CS

34 26112015

Interest in Crowdsourcing

Sara Fabrikant

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

humanmdashsystem (visuo-spatial) interface

spatio-temporal analytics

spatialization human navigation

multivariate spatial analysis etc)

interface design

large amp small interactive (map) displays

(ie desktop mobile VR etc)

empirical evaluations

(ie eye tracking EDA EEG etc) [httpwwwunitsevenconz]

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

human sensing observatory set up

controlled laboratory | in-situ (messy) real world setting

semi|automatic data collection of various human behavioral data streams

ie psycho-physiological perceptual cognitive and sensory-motor behaviors (through crowdsourcing )

Brian Fisher

My double research life

Brian Fisher

bull PhD in Experimental Psych

bull Postdoc w Cognitive Science

society founder amp president

bull Psychonomics Fellow

bull VIS-related symposia at Cogsci

amp APS papers on cogsci of

interaction

bull Fuzzy-logicBayes models

bull Postdoc funded by Inst for Robotics

and AI

bull VAST SC VEC VACCINE

bull Co-organized Dagstuhl ldquoInteraction

with Information for Visual Reasoningrdquo

Cogsci-based papers at VIS CHI

BELIV

My Plan

VIS

Me Psychological Science

Ujwal Gadiraju

p = 0055111

Crowd Worker

Behavior

Microtask

Design

Crowdsourcing Paradigm Quality amp Effectiveness

UJWAL GADIRAJU

Incentivization Gamification

Neha Gupta

Getting to know the crowd

Presented at Dagstuhl by

Dr David Martin amp Neha Gupta

Neha Gupta

PhD Student School of Computer Science University of Nottingham UK

Investigating crowdwork through the platform Amazon Mechanical Turk

psxng1nottinghamacuk

Neha_Gi_ji

Research amp Research teams

Research Studies

bull Qualitative study of Turkers on online forums like mturkforum and Turker Nation by Dr David Martin XRCE

bull Ethnographic Studies of Turkers in their respective homes and places of work in India by Neha Gupta UoN

Research teams

Xerox

bull David Martin Jacki OrsquoNeill (Now MSR India) Ben Hanrahan (Now PSU)

psxng1nottinghamacuk

Neha_Gi_ji

Nottingham

bull Andy Crabtree

Tom Rodden

Matthias Hirth

Matthias Hirth 48 48

About Myself

Matthias Hirth

Research Assistant at University of Wuumlrzburg

httpmatthias-hirthcom

matthiashirthinformatikuni-wuerzburgde

Crowdsourcing

Platform analysis

Platform optimization

Human factors

Use cases

Online Social Networks

Collaboration networks

Structural differences of OSNs

Application of OSN relationships

Interactive applicationsGaming

Challenges of real time interactions

Serious gamingGamification

E-Sports broadcastinglive streaming

Quality of Experience

Monitoring techniques

Quantification techniques

(Crowd-bases) test design

Social interactions

Test methodology Incentive design

Use case

Measurement technique

Identification of

qualification

Matthias Hirth 49 49

Improving Crowdsourcing Technology and the Way We Use it

Crowdsourcing technology

How to transfer well known lab procedures into an uncontrolled

crowdsourcing environment

Which crowdsourcing platform features are required to conduct

successful experiments but are not present yet

Which technical developments (can) foster the usage of

crowdsourcing instead of lab experiments

How can technological developments be used for novel incentive

mechanisms

Usage of Crowdsourcing

How can we conduct reproducible experiments in diverse

crowdsourcing environment

How can we raise the awareness of crowdsourcing employers for the

individuals behind the crowd

Tobias Hoszligfeld

The crowd community in the lab

Tobias Hoszligfeld wwwmaswiwiuni-duede

bull Background Tobias Hoszligfeld

ndash 2004-2014 Research group ldquoFuture Internet Applicationsrdquo Chair of Communication Networks Univ of Wuumlrzburg

ndash Since Sep 2014 Chair of Modeling of Adaptive Systems (head) University of Duisburg-Essen

bull Interests Application of crowdsourcing

ndash User studies on Quality of Experience (QoE) for Internet apps eg video QoE in the crowd amp in the lab methodology reliability statisticshellip [2]

ndash Mobile crowdsourcing network and context measurements eg Wifi offloading incentives privacy recommendation mechanisms etc [1]

bull Interests The crowd community

ndash Crowd = Human beings Social network wisdom of the crowd community

ndash UnderstandingUtilizing the crowd community for successful human computation For which jobs can social interaction improve the quality of work Classification

ndash How can we deploy such a system where interaction is triggered

ndash System design Task design interaction mechanisms incentives

22 Nov 2015

[1] httpdropsdagstuhldeopusvolltexte20134354

[2] httpswww3informatikuni-wuerzburgdeqoewikiqualinetcrowd

Challenges Example of Crowdsourced QoE

Tobias Hoszligfeld wwwmaswiwiuni-duede

task design

incentives

Internet

crowd

community

anonymous

remote

subjects

Internet delivery of

test data and user

response

QoE study bull testing methodology bull QoE influence factors

under investigation bull additional QoE factors bull proper subset of items bull hellip

crowdsourcing setting bull short task duration bull instructions amp training bull feedback amp questions bull hidden influence factors bull reliability of users bull hellip

heterogeneous devices

and Internet access

implementation extrinsic vs intrinsic

motivation

Processing of data

Technical aspects bull server implementation bull client application amp

compatibility bull desired test conditions bull context monitoring bull hellip

test items

lab

scientific methodologies

22 Nov 2015

Jason Jacques

Jason Jacques supervised by Per Ola Kristensson

T

Crowdsourcing Code Perception

jtj21camacuk peopledscamacukjtj21

jtjacquesgmailcom jsonjcouk

PhD Working Title

science

Crowdsourcing Interests W

hat

W

ho

Ho

w

Wh

y jtj21camacuk jsonjcouk science

Radu Jianu

Radu Jianu Assistant Professor

Florida International University (PhD Brown University lsquo12)

Interests

Applications of eye-tracking in visualization tracking what people look at rather than where they look

bull Data intensive large scale eye-tracking experiments bull Gaze-contingent visualizations bull Gaze-driven data recommendation and big data curation

Automating user studies in data visualizations (on next page)

Automating user studies

1 Software designs that can (semi)automate a broad range of quantitative evaluations by leveraging

ndash Crowdsourcing ndash Benchmark data and task taxonomies ndash Automated result interpretation

2 Moving user evaluation from the end to within the software design and implementation process as a means to choose between competitive visual designs

Christian Keimel

copy IRT

Where

What

bull Hybrid broadcast broadband technologies

bull Big Data in broadcasting environment and content-creation

bull Audio-visual QoE assessment (with crowdsourcing)

About me

Christian Keimel Slide 60

Lecturer for Digital Broadcast Engineering

Research Engineer for Media Services and Platforms

copy IRT

Crowdsourcing in large scale optimisation

The human component of Big Data

Moving crowdsourcing to lean-back devices

Crowdsourcing from the couch

Crowdsourced experiments emulating the laboratory

Is it time for new methodologies ignoring the lab

Questionshellip

Christian Keimel Slide 61

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 19: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

I have come here to learn

bull While I do a lot of empirical research most is qualitative

bull While I think that crowd sourcing is potentially a fantastic new resource I think successful use of crowd sourcing is non-trivial

bull I hope to learn about how to work with this

Remco Chang

21 Remco Chang ndash Dagstuhl 15

From vision science to data science applying perception to problems in big data

Remco Chang

Assistant Professor Computer Science

Tufts University

22 Remco Chang ndash Dagstuhl 15

Weberrsquos Law

119889119875 = 119896119889119878

119878

Min Chen

Min Chen Professor of Scientific Visualization

University of Oxford (2011 ~ Present) Swansea University (1984 ~ 2011)

Research Interests most aspects of visualization aspects of HCI computer graphics computer vision data mining and information theory

Research Highlight

1 Volume Graphics and Visualization

2 Video Visualization

3 Visual Analytics

4 Perception and Cognition in Visualization

5 Theory of Visualization

1 2

3 4

5

Research Question in the Context of 15481

Short Version How can we use crowdsourcing to capture perceptual and cognitive measures to facilitate mathematical modelling of visualization and interaction

Long Version Information theory is perhaps the most promising mathematical framework that can underpin the fields of visualization and interaction as well as data processing and communication However to obtain its core measures various types of entropy one needs to capture a probability distribution of all states associated with a concept While such a probability distribution can be captured in a machine-centric data processing or communication pipeline it is not trivial to capture in a human-centric process Crowdsourcing offers an opportunity to conduct ldquobigrdquo observation and measurement of perceptual and cognitive activities It is thus a tool for ldquoBig Sciencerdquo

Stephan Diehl

Dagstuhl Seminar 15481 Evaluation in the Crowd Crowdsourcing and Human-Centred Experiments

Stephan Diehl Software Engineering University of Trier Germany

The Java Code Clone

Detection API

Empirical Software Engineering bull [JSS2015 ESEM2015 FSE2014 ICSM 2012

FSE 2011 ESEM 2011 hellip]

Empirical Information Visualization bull [InfoVis 2010 J of Media Psychology 410 hellip]

Analysis and Collaboration SE Tools bull [ASE2014 ASE 2012 ICSE 2011 TSE 2005 ]

Visualizing (Dynamic) Compound Graphs bull [EuroVisSTAR 2014 IV Journal 2012 InfoVis11

SOFTVIS10 EuroVis09 IV09 EuroVis08

CHASE08 AVI08GD04GD02 hellip]

Dagstuhl Seminar 15481 Evaluation in the Crowd Crowdsourcing and Human-Centred Experiments

bull How to reach your (very specific) target audience without

spamming

bull How to control for resolution animation speed color

bull What crowdsourcing platforms exist and what are their

features

Darren Edwards

Darren Edwards Swansea University

Cognitive psychology - categorization

Health and clinical psychology ndash applications of categorization on clinical populations (eg autism)

Crowdsourcing and psychological interventions

Recent validity studies confirm the use of simple psychological tasks such as decision making

Recent attempts have now explored more complex tasks such as psychological interventions (eg mindfulness)

Present studies now explore more complex forms of mindfulness

What are the limitations of using crowdsourced populations online such as when using psychological interventions or categorization

Sebastian Egger

Engineering

Technical high school

bull Telematics

electrotechnical enginieering telecommunications

bull Signal Processing for UWB systems

Sociology

Theory of human societies

Interaction amp collaboration behaviour of groups

Measurement of human behaviour

Currently Human Computer Interaction

Social amp collaborative aspects of technology use

33 26112015

Sebastian Egger

Motivational aspects of crowdworkers

CS as a collaborative work setting

Optimal Setup of CS experiments

Questions (max Nr of questions max duration)

Gamified reliability checks

Scale designs

CS vs Lab whatlsquos the ground truth

New application domains of CS

34 26112015

Interest in Crowdsourcing

Sara Fabrikant

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

humanmdashsystem (visuo-spatial) interface

spatio-temporal analytics

spatialization human navigation

multivariate spatial analysis etc)

interface design

large amp small interactive (map) displays

(ie desktop mobile VR etc)

empirical evaluations

(ie eye tracking EDA EEG etc) [httpwwwunitsevenconz]

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

human sensing observatory set up

controlled laboratory | in-situ (messy) real world setting

semi|automatic data collection of various human behavioral data streams

ie psycho-physiological perceptual cognitive and sensory-motor behaviors (through crowdsourcing )

Brian Fisher

My double research life

Brian Fisher

bull PhD in Experimental Psych

bull Postdoc w Cognitive Science

society founder amp president

bull Psychonomics Fellow

bull VIS-related symposia at Cogsci

amp APS papers on cogsci of

interaction

bull Fuzzy-logicBayes models

bull Postdoc funded by Inst for Robotics

and AI

bull VAST SC VEC VACCINE

bull Co-organized Dagstuhl ldquoInteraction

with Information for Visual Reasoningrdquo

Cogsci-based papers at VIS CHI

BELIV

My Plan

VIS

Me Psychological Science

Ujwal Gadiraju

p = 0055111

Crowd Worker

Behavior

Microtask

Design

Crowdsourcing Paradigm Quality amp Effectiveness

UJWAL GADIRAJU

Incentivization Gamification

Neha Gupta

Getting to know the crowd

Presented at Dagstuhl by

Dr David Martin amp Neha Gupta

Neha Gupta

PhD Student School of Computer Science University of Nottingham UK

Investigating crowdwork through the platform Amazon Mechanical Turk

psxng1nottinghamacuk

Neha_Gi_ji

Research amp Research teams

Research Studies

bull Qualitative study of Turkers on online forums like mturkforum and Turker Nation by Dr David Martin XRCE

bull Ethnographic Studies of Turkers in their respective homes and places of work in India by Neha Gupta UoN

Research teams

Xerox

bull David Martin Jacki OrsquoNeill (Now MSR India) Ben Hanrahan (Now PSU)

psxng1nottinghamacuk

Neha_Gi_ji

Nottingham

bull Andy Crabtree

Tom Rodden

Matthias Hirth

Matthias Hirth 48 48

About Myself

Matthias Hirth

Research Assistant at University of Wuumlrzburg

httpmatthias-hirthcom

matthiashirthinformatikuni-wuerzburgde

Crowdsourcing

Platform analysis

Platform optimization

Human factors

Use cases

Online Social Networks

Collaboration networks

Structural differences of OSNs

Application of OSN relationships

Interactive applicationsGaming

Challenges of real time interactions

Serious gamingGamification

E-Sports broadcastinglive streaming

Quality of Experience

Monitoring techniques

Quantification techniques

(Crowd-bases) test design

Social interactions

Test methodology Incentive design

Use case

Measurement technique

Identification of

qualification

Matthias Hirth 49 49

Improving Crowdsourcing Technology and the Way We Use it

Crowdsourcing technology

How to transfer well known lab procedures into an uncontrolled

crowdsourcing environment

Which crowdsourcing platform features are required to conduct

successful experiments but are not present yet

Which technical developments (can) foster the usage of

crowdsourcing instead of lab experiments

How can technological developments be used for novel incentive

mechanisms

Usage of Crowdsourcing

How can we conduct reproducible experiments in diverse

crowdsourcing environment

How can we raise the awareness of crowdsourcing employers for the

individuals behind the crowd

Tobias Hoszligfeld

The crowd community in the lab

Tobias Hoszligfeld wwwmaswiwiuni-duede

bull Background Tobias Hoszligfeld

ndash 2004-2014 Research group ldquoFuture Internet Applicationsrdquo Chair of Communication Networks Univ of Wuumlrzburg

ndash Since Sep 2014 Chair of Modeling of Adaptive Systems (head) University of Duisburg-Essen

bull Interests Application of crowdsourcing

ndash User studies on Quality of Experience (QoE) for Internet apps eg video QoE in the crowd amp in the lab methodology reliability statisticshellip [2]

ndash Mobile crowdsourcing network and context measurements eg Wifi offloading incentives privacy recommendation mechanisms etc [1]

bull Interests The crowd community

ndash Crowd = Human beings Social network wisdom of the crowd community

ndash UnderstandingUtilizing the crowd community for successful human computation For which jobs can social interaction improve the quality of work Classification

ndash How can we deploy such a system where interaction is triggered

ndash System design Task design interaction mechanisms incentives

22 Nov 2015

[1] httpdropsdagstuhldeopusvolltexte20134354

[2] httpswww3informatikuni-wuerzburgdeqoewikiqualinetcrowd

Challenges Example of Crowdsourced QoE

Tobias Hoszligfeld wwwmaswiwiuni-duede

task design

incentives

Internet

crowd

community

anonymous

remote

subjects

Internet delivery of

test data and user

response

QoE study bull testing methodology bull QoE influence factors

under investigation bull additional QoE factors bull proper subset of items bull hellip

crowdsourcing setting bull short task duration bull instructions amp training bull feedback amp questions bull hidden influence factors bull reliability of users bull hellip

heterogeneous devices

and Internet access

implementation extrinsic vs intrinsic

motivation

Processing of data

Technical aspects bull server implementation bull client application amp

compatibility bull desired test conditions bull context monitoring bull hellip

test items

lab

scientific methodologies

22 Nov 2015

Jason Jacques

Jason Jacques supervised by Per Ola Kristensson

T

Crowdsourcing Code Perception

jtj21camacuk peopledscamacukjtj21

jtjacquesgmailcom jsonjcouk

PhD Working Title

science

Crowdsourcing Interests W

hat

W

ho

Ho

w

Wh

y jtj21camacuk jsonjcouk science

Radu Jianu

Radu Jianu Assistant Professor

Florida International University (PhD Brown University lsquo12)

Interests

Applications of eye-tracking in visualization tracking what people look at rather than where they look

bull Data intensive large scale eye-tracking experiments bull Gaze-contingent visualizations bull Gaze-driven data recommendation and big data curation

Automating user studies in data visualizations (on next page)

Automating user studies

1 Software designs that can (semi)automate a broad range of quantitative evaluations by leveraging

ndash Crowdsourcing ndash Benchmark data and task taxonomies ndash Automated result interpretation

2 Moving user evaluation from the end to within the software design and implementation process as a means to choose between competitive visual designs

Christian Keimel

copy IRT

Where

What

bull Hybrid broadcast broadband technologies

bull Big Data in broadcasting environment and content-creation

bull Audio-visual QoE assessment (with crowdsourcing)

About me

Christian Keimel Slide 60

Lecturer for Digital Broadcast Engineering

Research Engineer for Media Services and Platforms

copy IRT

Crowdsourcing in large scale optimisation

The human component of Big Data

Moving crowdsourcing to lean-back devices

Crowdsourcing from the couch

Crowdsourced experiments emulating the laboratory

Is it time for new methodologies ignoring the lab

Questionshellip

Christian Keimel Slide 61

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 20: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Remco Chang

21 Remco Chang ndash Dagstuhl 15

From vision science to data science applying perception to problems in big data

Remco Chang

Assistant Professor Computer Science

Tufts University

22 Remco Chang ndash Dagstuhl 15

Weberrsquos Law

119889119875 = 119896119889119878

119878

Min Chen

Min Chen Professor of Scientific Visualization

University of Oxford (2011 ~ Present) Swansea University (1984 ~ 2011)

Research Interests most aspects of visualization aspects of HCI computer graphics computer vision data mining and information theory

Research Highlight

1 Volume Graphics and Visualization

2 Video Visualization

3 Visual Analytics

4 Perception and Cognition in Visualization

5 Theory of Visualization

1 2

3 4

5

Research Question in the Context of 15481

Short Version How can we use crowdsourcing to capture perceptual and cognitive measures to facilitate mathematical modelling of visualization and interaction

Long Version Information theory is perhaps the most promising mathematical framework that can underpin the fields of visualization and interaction as well as data processing and communication However to obtain its core measures various types of entropy one needs to capture a probability distribution of all states associated with a concept While such a probability distribution can be captured in a machine-centric data processing or communication pipeline it is not trivial to capture in a human-centric process Crowdsourcing offers an opportunity to conduct ldquobigrdquo observation and measurement of perceptual and cognitive activities It is thus a tool for ldquoBig Sciencerdquo

Stephan Diehl

Dagstuhl Seminar 15481 Evaluation in the Crowd Crowdsourcing and Human-Centred Experiments

Stephan Diehl Software Engineering University of Trier Germany

The Java Code Clone

Detection API

Empirical Software Engineering bull [JSS2015 ESEM2015 FSE2014 ICSM 2012

FSE 2011 ESEM 2011 hellip]

Empirical Information Visualization bull [InfoVis 2010 J of Media Psychology 410 hellip]

Analysis and Collaboration SE Tools bull [ASE2014 ASE 2012 ICSE 2011 TSE 2005 ]

Visualizing (Dynamic) Compound Graphs bull [EuroVisSTAR 2014 IV Journal 2012 InfoVis11

SOFTVIS10 EuroVis09 IV09 EuroVis08

CHASE08 AVI08GD04GD02 hellip]

Dagstuhl Seminar 15481 Evaluation in the Crowd Crowdsourcing and Human-Centred Experiments

bull How to reach your (very specific) target audience without

spamming

bull How to control for resolution animation speed color

bull What crowdsourcing platforms exist and what are their

features

Darren Edwards

Darren Edwards Swansea University

Cognitive psychology - categorization

Health and clinical psychology ndash applications of categorization on clinical populations (eg autism)

Crowdsourcing and psychological interventions

Recent validity studies confirm the use of simple psychological tasks such as decision making

Recent attempts have now explored more complex tasks such as psychological interventions (eg mindfulness)

Present studies now explore more complex forms of mindfulness

What are the limitations of using crowdsourced populations online such as when using psychological interventions or categorization

Sebastian Egger

Engineering

Technical high school

bull Telematics

electrotechnical enginieering telecommunications

bull Signal Processing for UWB systems

Sociology

Theory of human societies

Interaction amp collaboration behaviour of groups

Measurement of human behaviour

Currently Human Computer Interaction

Social amp collaborative aspects of technology use

33 26112015

Sebastian Egger

Motivational aspects of crowdworkers

CS as a collaborative work setting

Optimal Setup of CS experiments

Questions (max Nr of questions max duration)

Gamified reliability checks

Scale designs

CS vs Lab whatlsquos the ground truth

New application domains of CS

34 26112015

Interest in Crowdsourcing

Sara Fabrikant

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

humanmdashsystem (visuo-spatial) interface

spatio-temporal analytics

spatialization human navigation

multivariate spatial analysis etc)

interface design

large amp small interactive (map) displays

(ie desktop mobile VR etc)

empirical evaluations

(ie eye tracking EDA EEG etc) [httpwwwunitsevenconz]

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

human sensing observatory set up

controlled laboratory | in-situ (messy) real world setting

semi|automatic data collection of various human behavioral data streams

ie psycho-physiological perceptual cognitive and sensory-motor behaviors (through crowdsourcing )

Brian Fisher

My double research life

Brian Fisher

bull PhD in Experimental Psych

bull Postdoc w Cognitive Science

society founder amp president

bull Psychonomics Fellow

bull VIS-related symposia at Cogsci

amp APS papers on cogsci of

interaction

bull Fuzzy-logicBayes models

bull Postdoc funded by Inst for Robotics

and AI

bull VAST SC VEC VACCINE

bull Co-organized Dagstuhl ldquoInteraction

with Information for Visual Reasoningrdquo

Cogsci-based papers at VIS CHI

BELIV

My Plan

VIS

Me Psychological Science

Ujwal Gadiraju

p = 0055111

Crowd Worker

Behavior

Microtask

Design

Crowdsourcing Paradigm Quality amp Effectiveness

UJWAL GADIRAJU

Incentivization Gamification

Neha Gupta

Getting to know the crowd

Presented at Dagstuhl by

Dr David Martin amp Neha Gupta

Neha Gupta

PhD Student School of Computer Science University of Nottingham UK

Investigating crowdwork through the platform Amazon Mechanical Turk

psxng1nottinghamacuk

Neha_Gi_ji

Research amp Research teams

Research Studies

bull Qualitative study of Turkers on online forums like mturkforum and Turker Nation by Dr David Martin XRCE

bull Ethnographic Studies of Turkers in their respective homes and places of work in India by Neha Gupta UoN

Research teams

Xerox

bull David Martin Jacki OrsquoNeill (Now MSR India) Ben Hanrahan (Now PSU)

psxng1nottinghamacuk

Neha_Gi_ji

Nottingham

bull Andy Crabtree

Tom Rodden

Matthias Hirth

Matthias Hirth 48 48

About Myself

Matthias Hirth

Research Assistant at University of Wuumlrzburg

httpmatthias-hirthcom

matthiashirthinformatikuni-wuerzburgde

Crowdsourcing

Platform analysis

Platform optimization

Human factors

Use cases

Online Social Networks

Collaboration networks

Structural differences of OSNs

Application of OSN relationships

Interactive applicationsGaming

Challenges of real time interactions

Serious gamingGamification

E-Sports broadcastinglive streaming

Quality of Experience

Monitoring techniques

Quantification techniques

(Crowd-bases) test design

Social interactions

Test methodology Incentive design

Use case

Measurement technique

Identification of

qualification

Matthias Hirth 49 49

Improving Crowdsourcing Technology and the Way We Use it

Crowdsourcing technology

How to transfer well known lab procedures into an uncontrolled

crowdsourcing environment

Which crowdsourcing platform features are required to conduct

successful experiments but are not present yet

Which technical developments (can) foster the usage of

crowdsourcing instead of lab experiments

How can technological developments be used for novel incentive

mechanisms

Usage of Crowdsourcing

How can we conduct reproducible experiments in diverse

crowdsourcing environment

How can we raise the awareness of crowdsourcing employers for the

individuals behind the crowd

Tobias Hoszligfeld

The crowd community in the lab

Tobias Hoszligfeld wwwmaswiwiuni-duede

bull Background Tobias Hoszligfeld

ndash 2004-2014 Research group ldquoFuture Internet Applicationsrdquo Chair of Communication Networks Univ of Wuumlrzburg

ndash Since Sep 2014 Chair of Modeling of Adaptive Systems (head) University of Duisburg-Essen

bull Interests Application of crowdsourcing

ndash User studies on Quality of Experience (QoE) for Internet apps eg video QoE in the crowd amp in the lab methodology reliability statisticshellip [2]

ndash Mobile crowdsourcing network and context measurements eg Wifi offloading incentives privacy recommendation mechanisms etc [1]

bull Interests The crowd community

ndash Crowd = Human beings Social network wisdom of the crowd community

ndash UnderstandingUtilizing the crowd community for successful human computation For which jobs can social interaction improve the quality of work Classification

ndash How can we deploy such a system where interaction is triggered

ndash System design Task design interaction mechanisms incentives

22 Nov 2015

[1] httpdropsdagstuhldeopusvolltexte20134354

[2] httpswww3informatikuni-wuerzburgdeqoewikiqualinetcrowd

Challenges Example of Crowdsourced QoE

Tobias Hoszligfeld wwwmaswiwiuni-duede

task design

incentives

Internet

crowd

community

anonymous

remote

subjects

Internet delivery of

test data and user

response

QoE study bull testing methodology bull QoE influence factors

under investigation bull additional QoE factors bull proper subset of items bull hellip

crowdsourcing setting bull short task duration bull instructions amp training bull feedback amp questions bull hidden influence factors bull reliability of users bull hellip

heterogeneous devices

and Internet access

implementation extrinsic vs intrinsic

motivation

Processing of data

Technical aspects bull server implementation bull client application amp

compatibility bull desired test conditions bull context monitoring bull hellip

test items

lab

scientific methodologies

22 Nov 2015

Jason Jacques

Jason Jacques supervised by Per Ola Kristensson

T

Crowdsourcing Code Perception

jtj21camacuk peopledscamacukjtj21

jtjacquesgmailcom jsonjcouk

PhD Working Title

science

Crowdsourcing Interests W

hat

W

ho

Ho

w

Wh

y jtj21camacuk jsonjcouk science

Radu Jianu

Radu Jianu Assistant Professor

Florida International University (PhD Brown University lsquo12)

Interests

Applications of eye-tracking in visualization tracking what people look at rather than where they look

bull Data intensive large scale eye-tracking experiments bull Gaze-contingent visualizations bull Gaze-driven data recommendation and big data curation

Automating user studies in data visualizations (on next page)

Automating user studies

1 Software designs that can (semi)automate a broad range of quantitative evaluations by leveraging

ndash Crowdsourcing ndash Benchmark data and task taxonomies ndash Automated result interpretation

2 Moving user evaluation from the end to within the software design and implementation process as a means to choose between competitive visual designs

Christian Keimel

copy IRT

Where

What

bull Hybrid broadcast broadband technologies

bull Big Data in broadcasting environment and content-creation

bull Audio-visual QoE assessment (with crowdsourcing)

About me

Christian Keimel Slide 60

Lecturer for Digital Broadcast Engineering

Research Engineer for Media Services and Platforms

copy IRT

Crowdsourcing in large scale optimisation

The human component of Big Data

Moving crowdsourcing to lean-back devices

Crowdsourcing from the couch

Crowdsourced experiments emulating the laboratory

Is it time for new methodologies ignoring the lab

Questionshellip

Christian Keimel Slide 61

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 21: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

21 Remco Chang ndash Dagstuhl 15

From vision science to data science applying perception to problems in big data

Remco Chang

Assistant Professor Computer Science

Tufts University

22 Remco Chang ndash Dagstuhl 15

Weberrsquos Law

119889119875 = 119896119889119878

119878

Min Chen

Min Chen Professor of Scientific Visualization

University of Oxford (2011 ~ Present) Swansea University (1984 ~ 2011)

Research Interests most aspects of visualization aspects of HCI computer graphics computer vision data mining and information theory

Research Highlight

1 Volume Graphics and Visualization

2 Video Visualization

3 Visual Analytics

4 Perception and Cognition in Visualization

5 Theory of Visualization

1 2

3 4

5

Research Question in the Context of 15481

Short Version How can we use crowdsourcing to capture perceptual and cognitive measures to facilitate mathematical modelling of visualization and interaction

Long Version Information theory is perhaps the most promising mathematical framework that can underpin the fields of visualization and interaction as well as data processing and communication However to obtain its core measures various types of entropy one needs to capture a probability distribution of all states associated with a concept While such a probability distribution can be captured in a machine-centric data processing or communication pipeline it is not trivial to capture in a human-centric process Crowdsourcing offers an opportunity to conduct ldquobigrdquo observation and measurement of perceptual and cognitive activities It is thus a tool for ldquoBig Sciencerdquo

Stephan Diehl

Dagstuhl Seminar 15481 Evaluation in the Crowd Crowdsourcing and Human-Centred Experiments

Stephan Diehl Software Engineering University of Trier Germany

The Java Code Clone

Detection API

Empirical Software Engineering bull [JSS2015 ESEM2015 FSE2014 ICSM 2012

FSE 2011 ESEM 2011 hellip]

Empirical Information Visualization bull [InfoVis 2010 J of Media Psychology 410 hellip]

Analysis and Collaboration SE Tools bull [ASE2014 ASE 2012 ICSE 2011 TSE 2005 ]

Visualizing (Dynamic) Compound Graphs bull [EuroVisSTAR 2014 IV Journal 2012 InfoVis11

SOFTVIS10 EuroVis09 IV09 EuroVis08

CHASE08 AVI08GD04GD02 hellip]

Dagstuhl Seminar 15481 Evaluation in the Crowd Crowdsourcing and Human-Centred Experiments

bull How to reach your (very specific) target audience without

spamming

bull How to control for resolution animation speed color

bull What crowdsourcing platforms exist and what are their

features

Darren Edwards

Darren Edwards Swansea University

Cognitive psychology - categorization

Health and clinical psychology ndash applications of categorization on clinical populations (eg autism)

Crowdsourcing and psychological interventions

Recent validity studies confirm the use of simple psychological tasks such as decision making

Recent attempts have now explored more complex tasks such as psychological interventions (eg mindfulness)

Present studies now explore more complex forms of mindfulness

What are the limitations of using crowdsourced populations online such as when using psychological interventions or categorization

Sebastian Egger

Engineering

Technical high school

bull Telematics

electrotechnical enginieering telecommunications

bull Signal Processing for UWB systems

Sociology

Theory of human societies

Interaction amp collaboration behaviour of groups

Measurement of human behaviour

Currently Human Computer Interaction

Social amp collaborative aspects of technology use

33 26112015

Sebastian Egger

Motivational aspects of crowdworkers

CS as a collaborative work setting

Optimal Setup of CS experiments

Questions (max Nr of questions max duration)

Gamified reliability checks

Scale designs

CS vs Lab whatlsquos the ground truth

New application domains of CS

34 26112015

Interest in Crowdsourcing

Sara Fabrikant

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

humanmdashsystem (visuo-spatial) interface

spatio-temporal analytics

spatialization human navigation

multivariate spatial analysis etc)

interface design

large amp small interactive (map) displays

(ie desktop mobile VR etc)

empirical evaluations

(ie eye tracking EDA EEG etc) [httpwwwunitsevenconz]

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

human sensing observatory set up

controlled laboratory | in-situ (messy) real world setting

semi|automatic data collection of various human behavioral data streams

ie psycho-physiological perceptual cognitive and sensory-motor behaviors (through crowdsourcing )

Brian Fisher

My double research life

Brian Fisher

bull PhD in Experimental Psych

bull Postdoc w Cognitive Science

society founder amp president

bull Psychonomics Fellow

bull VIS-related symposia at Cogsci

amp APS papers on cogsci of

interaction

bull Fuzzy-logicBayes models

bull Postdoc funded by Inst for Robotics

and AI

bull VAST SC VEC VACCINE

bull Co-organized Dagstuhl ldquoInteraction

with Information for Visual Reasoningrdquo

Cogsci-based papers at VIS CHI

BELIV

My Plan

VIS

Me Psychological Science

Ujwal Gadiraju

p = 0055111

Crowd Worker

Behavior

Microtask

Design

Crowdsourcing Paradigm Quality amp Effectiveness

UJWAL GADIRAJU

Incentivization Gamification

Neha Gupta

Getting to know the crowd

Presented at Dagstuhl by

Dr David Martin amp Neha Gupta

Neha Gupta

PhD Student School of Computer Science University of Nottingham UK

Investigating crowdwork through the platform Amazon Mechanical Turk

psxng1nottinghamacuk

Neha_Gi_ji

Research amp Research teams

Research Studies

bull Qualitative study of Turkers on online forums like mturkforum and Turker Nation by Dr David Martin XRCE

bull Ethnographic Studies of Turkers in their respective homes and places of work in India by Neha Gupta UoN

Research teams

Xerox

bull David Martin Jacki OrsquoNeill (Now MSR India) Ben Hanrahan (Now PSU)

psxng1nottinghamacuk

Neha_Gi_ji

Nottingham

bull Andy Crabtree

Tom Rodden

Matthias Hirth

Matthias Hirth 48 48

About Myself

Matthias Hirth

Research Assistant at University of Wuumlrzburg

httpmatthias-hirthcom

matthiashirthinformatikuni-wuerzburgde

Crowdsourcing

Platform analysis

Platform optimization

Human factors

Use cases

Online Social Networks

Collaboration networks

Structural differences of OSNs

Application of OSN relationships

Interactive applicationsGaming

Challenges of real time interactions

Serious gamingGamification

E-Sports broadcastinglive streaming

Quality of Experience

Monitoring techniques

Quantification techniques

(Crowd-bases) test design

Social interactions

Test methodology Incentive design

Use case

Measurement technique

Identification of

qualification

Matthias Hirth 49 49

Improving Crowdsourcing Technology and the Way We Use it

Crowdsourcing technology

How to transfer well known lab procedures into an uncontrolled

crowdsourcing environment

Which crowdsourcing platform features are required to conduct

successful experiments but are not present yet

Which technical developments (can) foster the usage of

crowdsourcing instead of lab experiments

How can technological developments be used for novel incentive

mechanisms

Usage of Crowdsourcing

How can we conduct reproducible experiments in diverse

crowdsourcing environment

How can we raise the awareness of crowdsourcing employers for the

individuals behind the crowd

Tobias Hoszligfeld

The crowd community in the lab

Tobias Hoszligfeld wwwmaswiwiuni-duede

bull Background Tobias Hoszligfeld

ndash 2004-2014 Research group ldquoFuture Internet Applicationsrdquo Chair of Communication Networks Univ of Wuumlrzburg

ndash Since Sep 2014 Chair of Modeling of Adaptive Systems (head) University of Duisburg-Essen

bull Interests Application of crowdsourcing

ndash User studies on Quality of Experience (QoE) for Internet apps eg video QoE in the crowd amp in the lab methodology reliability statisticshellip [2]

ndash Mobile crowdsourcing network and context measurements eg Wifi offloading incentives privacy recommendation mechanisms etc [1]

bull Interests The crowd community

ndash Crowd = Human beings Social network wisdom of the crowd community

ndash UnderstandingUtilizing the crowd community for successful human computation For which jobs can social interaction improve the quality of work Classification

ndash How can we deploy such a system where interaction is triggered

ndash System design Task design interaction mechanisms incentives

22 Nov 2015

[1] httpdropsdagstuhldeopusvolltexte20134354

[2] httpswww3informatikuni-wuerzburgdeqoewikiqualinetcrowd

Challenges Example of Crowdsourced QoE

Tobias Hoszligfeld wwwmaswiwiuni-duede

task design

incentives

Internet

crowd

community

anonymous

remote

subjects

Internet delivery of

test data and user

response

QoE study bull testing methodology bull QoE influence factors

under investigation bull additional QoE factors bull proper subset of items bull hellip

crowdsourcing setting bull short task duration bull instructions amp training bull feedback amp questions bull hidden influence factors bull reliability of users bull hellip

heterogeneous devices

and Internet access

implementation extrinsic vs intrinsic

motivation

Processing of data

Technical aspects bull server implementation bull client application amp

compatibility bull desired test conditions bull context monitoring bull hellip

test items

lab

scientific methodologies

22 Nov 2015

Jason Jacques

Jason Jacques supervised by Per Ola Kristensson

T

Crowdsourcing Code Perception

jtj21camacuk peopledscamacukjtj21

jtjacquesgmailcom jsonjcouk

PhD Working Title

science

Crowdsourcing Interests W

hat

W

ho

Ho

w

Wh

y jtj21camacuk jsonjcouk science

Radu Jianu

Radu Jianu Assistant Professor

Florida International University (PhD Brown University lsquo12)

Interests

Applications of eye-tracking in visualization tracking what people look at rather than where they look

bull Data intensive large scale eye-tracking experiments bull Gaze-contingent visualizations bull Gaze-driven data recommendation and big data curation

Automating user studies in data visualizations (on next page)

Automating user studies

1 Software designs that can (semi)automate a broad range of quantitative evaluations by leveraging

ndash Crowdsourcing ndash Benchmark data and task taxonomies ndash Automated result interpretation

2 Moving user evaluation from the end to within the software design and implementation process as a means to choose between competitive visual designs

Christian Keimel

copy IRT

Where

What

bull Hybrid broadcast broadband technologies

bull Big Data in broadcasting environment and content-creation

bull Audio-visual QoE assessment (with crowdsourcing)

About me

Christian Keimel Slide 60

Lecturer for Digital Broadcast Engineering

Research Engineer for Media Services and Platforms

copy IRT

Crowdsourcing in large scale optimisation

The human component of Big Data

Moving crowdsourcing to lean-back devices

Crowdsourcing from the couch

Crowdsourced experiments emulating the laboratory

Is it time for new methodologies ignoring the lab

Questionshellip

Christian Keimel Slide 61

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 22: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

22 Remco Chang ndash Dagstuhl 15

Weberrsquos Law

119889119875 = 119896119889119878

119878

Min Chen

Min Chen Professor of Scientific Visualization

University of Oxford (2011 ~ Present) Swansea University (1984 ~ 2011)

Research Interests most aspects of visualization aspects of HCI computer graphics computer vision data mining and information theory

Research Highlight

1 Volume Graphics and Visualization

2 Video Visualization

3 Visual Analytics

4 Perception and Cognition in Visualization

5 Theory of Visualization

1 2

3 4

5

Research Question in the Context of 15481

Short Version How can we use crowdsourcing to capture perceptual and cognitive measures to facilitate mathematical modelling of visualization and interaction

Long Version Information theory is perhaps the most promising mathematical framework that can underpin the fields of visualization and interaction as well as data processing and communication However to obtain its core measures various types of entropy one needs to capture a probability distribution of all states associated with a concept While such a probability distribution can be captured in a machine-centric data processing or communication pipeline it is not trivial to capture in a human-centric process Crowdsourcing offers an opportunity to conduct ldquobigrdquo observation and measurement of perceptual and cognitive activities It is thus a tool for ldquoBig Sciencerdquo

Stephan Diehl

Dagstuhl Seminar 15481 Evaluation in the Crowd Crowdsourcing and Human-Centred Experiments

Stephan Diehl Software Engineering University of Trier Germany

The Java Code Clone

Detection API

Empirical Software Engineering bull [JSS2015 ESEM2015 FSE2014 ICSM 2012

FSE 2011 ESEM 2011 hellip]

Empirical Information Visualization bull [InfoVis 2010 J of Media Psychology 410 hellip]

Analysis and Collaboration SE Tools bull [ASE2014 ASE 2012 ICSE 2011 TSE 2005 ]

Visualizing (Dynamic) Compound Graphs bull [EuroVisSTAR 2014 IV Journal 2012 InfoVis11

SOFTVIS10 EuroVis09 IV09 EuroVis08

CHASE08 AVI08GD04GD02 hellip]

Dagstuhl Seminar 15481 Evaluation in the Crowd Crowdsourcing and Human-Centred Experiments

bull How to reach your (very specific) target audience without

spamming

bull How to control for resolution animation speed color

bull What crowdsourcing platforms exist and what are their

features

Darren Edwards

Darren Edwards Swansea University

Cognitive psychology - categorization

Health and clinical psychology ndash applications of categorization on clinical populations (eg autism)

Crowdsourcing and psychological interventions

Recent validity studies confirm the use of simple psychological tasks such as decision making

Recent attempts have now explored more complex tasks such as psychological interventions (eg mindfulness)

Present studies now explore more complex forms of mindfulness

What are the limitations of using crowdsourced populations online such as when using psychological interventions or categorization

Sebastian Egger

Engineering

Technical high school

bull Telematics

electrotechnical enginieering telecommunications

bull Signal Processing for UWB systems

Sociology

Theory of human societies

Interaction amp collaboration behaviour of groups

Measurement of human behaviour

Currently Human Computer Interaction

Social amp collaborative aspects of technology use

33 26112015

Sebastian Egger

Motivational aspects of crowdworkers

CS as a collaborative work setting

Optimal Setup of CS experiments

Questions (max Nr of questions max duration)

Gamified reliability checks

Scale designs

CS vs Lab whatlsquos the ground truth

New application domains of CS

34 26112015

Interest in Crowdsourcing

Sara Fabrikant

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

humanmdashsystem (visuo-spatial) interface

spatio-temporal analytics

spatialization human navigation

multivariate spatial analysis etc)

interface design

large amp small interactive (map) displays

(ie desktop mobile VR etc)

empirical evaluations

(ie eye tracking EDA EEG etc) [httpwwwunitsevenconz]

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

human sensing observatory set up

controlled laboratory | in-situ (messy) real world setting

semi|automatic data collection of various human behavioral data streams

ie psycho-physiological perceptual cognitive and sensory-motor behaviors (through crowdsourcing )

Brian Fisher

My double research life

Brian Fisher

bull PhD in Experimental Psych

bull Postdoc w Cognitive Science

society founder amp president

bull Psychonomics Fellow

bull VIS-related symposia at Cogsci

amp APS papers on cogsci of

interaction

bull Fuzzy-logicBayes models

bull Postdoc funded by Inst for Robotics

and AI

bull VAST SC VEC VACCINE

bull Co-organized Dagstuhl ldquoInteraction

with Information for Visual Reasoningrdquo

Cogsci-based papers at VIS CHI

BELIV

My Plan

VIS

Me Psychological Science

Ujwal Gadiraju

p = 0055111

Crowd Worker

Behavior

Microtask

Design

Crowdsourcing Paradigm Quality amp Effectiveness

UJWAL GADIRAJU

Incentivization Gamification

Neha Gupta

Getting to know the crowd

Presented at Dagstuhl by

Dr David Martin amp Neha Gupta

Neha Gupta

PhD Student School of Computer Science University of Nottingham UK

Investigating crowdwork through the platform Amazon Mechanical Turk

psxng1nottinghamacuk

Neha_Gi_ji

Research amp Research teams

Research Studies

bull Qualitative study of Turkers on online forums like mturkforum and Turker Nation by Dr David Martin XRCE

bull Ethnographic Studies of Turkers in their respective homes and places of work in India by Neha Gupta UoN

Research teams

Xerox

bull David Martin Jacki OrsquoNeill (Now MSR India) Ben Hanrahan (Now PSU)

psxng1nottinghamacuk

Neha_Gi_ji

Nottingham

bull Andy Crabtree

Tom Rodden

Matthias Hirth

Matthias Hirth 48 48

About Myself

Matthias Hirth

Research Assistant at University of Wuumlrzburg

httpmatthias-hirthcom

matthiashirthinformatikuni-wuerzburgde

Crowdsourcing

Platform analysis

Platform optimization

Human factors

Use cases

Online Social Networks

Collaboration networks

Structural differences of OSNs

Application of OSN relationships

Interactive applicationsGaming

Challenges of real time interactions

Serious gamingGamification

E-Sports broadcastinglive streaming

Quality of Experience

Monitoring techniques

Quantification techniques

(Crowd-bases) test design

Social interactions

Test methodology Incentive design

Use case

Measurement technique

Identification of

qualification

Matthias Hirth 49 49

Improving Crowdsourcing Technology and the Way We Use it

Crowdsourcing technology

How to transfer well known lab procedures into an uncontrolled

crowdsourcing environment

Which crowdsourcing platform features are required to conduct

successful experiments but are not present yet

Which technical developments (can) foster the usage of

crowdsourcing instead of lab experiments

How can technological developments be used for novel incentive

mechanisms

Usage of Crowdsourcing

How can we conduct reproducible experiments in diverse

crowdsourcing environment

How can we raise the awareness of crowdsourcing employers for the

individuals behind the crowd

Tobias Hoszligfeld

The crowd community in the lab

Tobias Hoszligfeld wwwmaswiwiuni-duede

bull Background Tobias Hoszligfeld

ndash 2004-2014 Research group ldquoFuture Internet Applicationsrdquo Chair of Communication Networks Univ of Wuumlrzburg

ndash Since Sep 2014 Chair of Modeling of Adaptive Systems (head) University of Duisburg-Essen

bull Interests Application of crowdsourcing

ndash User studies on Quality of Experience (QoE) for Internet apps eg video QoE in the crowd amp in the lab methodology reliability statisticshellip [2]

ndash Mobile crowdsourcing network and context measurements eg Wifi offloading incentives privacy recommendation mechanisms etc [1]

bull Interests The crowd community

ndash Crowd = Human beings Social network wisdom of the crowd community

ndash UnderstandingUtilizing the crowd community for successful human computation For which jobs can social interaction improve the quality of work Classification

ndash How can we deploy such a system where interaction is triggered

ndash System design Task design interaction mechanisms incentives

22 Nov 2015

[1] httpdropsdagstuhldeopusvolltexte20134354

[2] httpswww3informatikuni-wuerzburgdeqoewikiqualinetcrowd

Challenges Example of Crowdsourced QoE

Tobias Hoszligfeld wwwmaswiwiuni-duede

task design

incentives

Internet

crowd

community

anonymous

remote

subjects

Internet delivery of

test data and user

response

QoE study bull testing methodology bull QoE influence factors

under investigation bull additional QoE factors bull proper subset of items bull hellip

crowdsourcing setting bull short task duration bull instructions amp training bull feedback amp questions bull hidden influence factors bull reliability of users bull hellip

heterogeneous devices

and Internet access

implementation extrinsic vs intrinsic

motivation

Processing of data

Technical aspects bull server implementation bull client application amp

compatibility bull desired test conditions bull context monitoring bull hellip

test items

lab

scientific methodologies

22 Nov 2015

Jason Jacques

Jason Jacques supervised by Per Ola Kristensson

T

Crowdsourcing Code Perception

jtj21camacuk peopledscamacukjtj21

jtjacquesgmailcom jsonjcouk

PhD Working Title

science

Crowdsourcing Interests W

hat

W

ho

Ho

w

Wh

y jtj21camacuk jsonjcouk science

Radu Jianu

Radu Jianu Assistant Professor

Florida International University (PhD Brown University lsquo12)

Interests

Applications of eye-tracking in visualization tracking what people look at rather than where they look

bull Data intensive large scale eye-tracking experiments bull Gaze-contingent visualizations bull Gaze-driven data recommendation and big data curation

Automating user studies in data visualizations (on next page)

Automating user studies

1 Software designs that can (semi)automate a broad range of quantitative evaluations by leveraging

ndash Crowdsourcing ndash Benchmark data and task taxonomies ndash Automated result interpretation

2 Moving user evaluation from the end to within the software design and implementation process as a means to choose between competitive visual designs

Christian Keimel

copy IRT

Where

What

bull Hybrid broadcast broadband technologies

bull Big Data in broadcasting environment and content-creation

bull Audio-visual QoE assessment (with crowdsourcing)

About me

Christian Keimel Slide 60

Lecturer for Digital Broadcast Engineering

Research Engineer for Media Services and Platforms

copy IRT

Crowdsourcing in large scale optimisation

The human component of Big Data

Moving crowdsourcing to lean-back devices

Crowdsourcing from the couch

Crowdsourced experiments emulating the laboratory

Is it time for new methodologies ignoring the lab

Questionshellip

Christian Keimel Slide 61

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 23: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Min Chen

Min Chen Professor of Scientific Visualization

University of Oxford (2011 ~ Present) Swansea University (1984 ~ 2011)

Research Interests most aspects of visualization aspects of HCI computer graphics computer vision data mining and information theory

Research Highlight

1 Volume Graphics and Visualization

2 Video Visualization

3 Visual Analytics

4 Perception and Cognition in Visualization

5 Theory of Visualization

1 2

3 4

5

Research Question in the Context of 15481

Short Version How can we use crowdsourcing to capture perceptual and cognitive measures to facilitate mathematical modelling of visualization and interaction

Long Version Information theory is perhaps the most promising mathematical framework that can underpin the fields of visualization and interaction as well as data processing and communication However to obtain its core measures various types of entropy one needs to capture a probability distribution of all states associated with a concept While such a probability distribution can be captured in a machine-centric data processing or communication pipeline it is not trivial to capture in a human-centric process Crowdsourcing offers an opportunity to conduct ldquobigrdquo observation and measurement of perceptual and cognitive activities It is thus a tool for ldquoBig Sciencerdquo

Stephan Diehl

Dagstuhl Seminar 15481 Evaluation in the Crowd Crowdsourcing and Human-Centred Experiments

Stephan Diehl Software Engineering University of Trier Germany

The Java Code Clone

Detection API

Empirical Software Engineering bull [JSS2015 ESEM2015 FSE2014 ICSM 2012

FSE 2011 ESEM 2011 hellip]

Empirical Information Visualization bull [InfoVis 2010 J of Media Psychology 410 hellip]

Analysis and Collaboration SE Tools bull [ASE2014 ASE 2012 ICSE 2011 TSE 2005 ]

Visualizing (Dynamic) Compound Graphs bull [EuroVisSTAR 2014 IV Journal 2012 InfoVis11

SOFTVIS10 EuroVis09 IV09 EuroVis08

CHASE08 AVI08GD04GD02 hellip]

Dagstuhl Seminar 15481 Evaluation in the Crowd Crowdsourcing and Human-Centred Experiments

bull How to reach your (very specific) target audience without

spamming

bull How to control for resolution animation speed color

bull What crowdsourcing platforms exist and what are their

features

Darren Edwards

Darren Edwards Swansea University

Cognitive psychology - categorization

Health and clinical psychology ndash applications of categorization on clinical populations (eg autism)

Crowdsourcing and psychological interventions

Recent validity studies confirm the use of simple psychological tasks such as decision making

Recent attempts have now explored more complex tasks such as psychological interventions (eg mindfulness)

Present studies now explore more complex forms of mindfulness

What are the limitations of using crowdsourced populations online such as when using psychological interventions or categorization

Sebastian Egger

Engineering

Technical high school

bull Telematics

electrotechnical enginieering telecommunications

bull Signal Processing for UWB systems

Sociology

Theory of human societies

Interaction amp collaboration behaviour of groups

Measurement of human behaviour

Currently Human Computer Interaction

Social amp collaborative aspects of technology use

33 26112015

Sebastian Egger

Motivational aspects of crowdworkers

CS as a collaborative work setting

Optimal Setup of CS experiments

Questions (max Nr of questions max duration)

Gamified reliability checks

Scale designs

CS vs Lab whatlsquos the ground truth

New application domains of CS

34 26112015

Interest in Crowdsourcing

Sara Fabrikant

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

humanmdashsystem (visuo-spatial) interface

spatio-temporal analytics

spatialization human navigation

multivariate spatial analysis etc)

interface design

large amp small interactive (map) displays

(ie desktop mobile VR etc)

empirical evaluations

(ie eye tracking EDA EEG etc) [httpwwwunitsevenconz]

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

human sensing observatory set up

controlled laboratory | in-situ (messy) real world setting

semi|automatic data collection of various human behavioral data streams

ie psycho-physiological perceptual cognitive and sensory-motor behaviors (through crowdsourcing )

Brian Fisher

My double research life

Brian Fisher

bull PhD in Experimental Psych

bull Postdoc w Cognitive Science

society founder amp president

bull Psychonomics Fellow

bull VIS-related symposia at Cogsci

amp APS papers on cogsci of

interaction

bull Fuzzy-logicBayes models

bull Postdoc funded by Inst for Robotics

and AI

bull VAST SC VEC VACCINE

bull Co-organized Dagstuhl ldquoInteraction

with Information for Visual Reasoningrdquo

Cogsci-based papers at VIS CHI

BELIV

My Plan

VIS

Me Psychological Science

Ujwal Gadiraju

p = 0055111

Crowd Worker

Behavior

Microtask

Design

Crowdsourcing Paradigm Quality amp Effectiveness

UJWAL GADIRAJU

Incentivization Gamification

Neha Gupta

Getting to know the crowd

Presented at Dagstuhl by

Dr David Martin amp Neha Gupta

Neha Gupta

PhD Student School of Computer Science University of Nottingham UK

Investigating crowdwork through the platform Amazon Mechanical Turk

psxng1nottinghamacuk

Neha_Gi_ji

Research amp Research teams

Research Studies

bull Qualitative study of Turkers on online forums like mturkforum and Turker Nation by Dr David Martin XRCE

bull Ethnographic Studies of Turkers in their respective homes and places of work in India by Neha Gupta UoN

Research teams

Xerox

bull David Martin Jacki OrsquoNeill (Now MSR India) Ben Hanrahan (Now PSU)

psxng1nottinghamacuk

Neha_Gi_ji

Nottingham

bull Andy Crabtree

Tom Rodden

Matthias Hirth

Matthias Hirth 48 48

About Myself

Matthias Hirth

Research Assistant at University of Wuumlrzburg

httpmatthias-hirthcom

matthiashirthinformatikuni-wuerzburgde

Crowdsourcing

Platform analysis

Platform optimization

Human factors

Use cases

Online Social Networks

Collaboration networks

Structural differences of OSNs

Application of OSN relationships

Interactive applicationsGaming

Challenges of real time interactions

Serious gamingGamification

E-Sports broadcastinglive streaming

Quality of Experience

Monitoring techniques

Quantification techniques

(Crowd-bases) test design

Social interactions

Test methodology Incentive design

Use case

Measurement technique

Identification of

qualification

Matthias Hirth 49 49

Improving Crowdsourcing Technology and the Way We Use it

Crowdsourcing technology

How to transfer well known lab procedures into an uncontrolled

crowdsourcing environment

Which crowdsourcing platform features are required to conduct

successful experiments but are not present yet

Which technical developments (can) foster the usage of

crowdsourcing instead of lab experiments

How can technological developments be used for novel incentive

mechanisms

Usage of Crowdsourcing

How can we conduct reproducible experiments in diverse

crowdsourcing environment

How can we raise the awareness of crowdsourcing employers for the

individuals behind the crowd

Tobias Hoszligfeld

The crowd community in the lab

Tobias Hoszligfeld wwwmaswiwiuni-duede

bull Background Tobias Hoszligfeld

ndash 2004-2014 Research group ldquoFuture Internet Applicationsrdquo Chair of Communication Networks Univ of Wuumlrzburg

ndash Since Sep 2014 Chair of Modeling of Adaptive Systems (head) University of Duisburg-Essen

bull Interests Application of crowdsourcing

ndash User studies on Quality of Experience (QoE) for Internet apps eg video QoE in the crowd amp in the lab methodology reliability statisticshellip [2]

ndash Mobile crowdsourcing network and context measurements eg Wifi offloading incentives privacy recommendation mechanisms etc [1]

bull Interests The crowd community

ndash Crowd = Human beings Social network wisdom of the crowd community

ndash UnderstandingUtilizing the crowd community for successful human computation For which jobs can social interaction improve the quality of work Classification

ndash How can we deploy such a system where interaction is triggered

ndash System design Task design interaction mechanisms incentives

22 Nov 2015

[1] httpdropsdagstuhldeopusvolltexte20134354

[2] httpswww3informatikuni-wuerzburgdeqoewikiqualinetcrowd

Challenges Example of Crowdsourced QoE

Tobias Hoszligfeld wwwmaswiwiuni-duede

task design

incentives

Internet

crowd

community

anonymous

remote

subjects

Internet delivery of

test data and user

response

QoE study bull testing methodology bull QoE influence factors

under investigation bull additional QoE factors bull proper subset of items bull hellip

crowdsourcing setting bull short task duration bull instructions amp training bull feedback amp questions bull hidden influence factors bull reliability of users bull hellip

heterogeneous devices

and Internet access

implementation extrinsic vs intrinsic

motivation

Processing of data

Technical aspects bull server implementation bull client application amp

compatibility bull desired test conditions bull context monitoring bull hellip

test items

lab

scientific methodologies

22 Nov 2015

Jason Jacques

Jason Jacques supervised by Per Ola Kristensson

T

Crowdsourcing Code Perception

jtj21camacuk peopledscamacukjtj21

jtjacquesgmailcom jsonjcouk

PhD Working Title

science

Crowdsourcing Interests W

hat

W

ho

Ho

w

Wh

y jtj21camacuk jsonjcouk science

Radu Jianu

Radu Jianu Assistant Professor

Florida International University (PhD Brown University lsquo12)

Interests

Applications of eye-tracking in visualization tracking what people look at rather than where they look

bull Data intensive large scale eye-tracking experiments bull Gaze-contingent visualizations bull Gaze-driven data recommendation and big data curation

Automating user studies in data visualizations (on next page)

Automating user studies

1 Software designs that can (semi)automate a broad range of quantitative evaluations by leveraging

ndash Crowdsourcing ndash Benchmark data and task taxonomies ndash Automated result interpretation

2 Moving user evaluation from the end to within the software design and implementation process as a means to choose between competitive visual designs

Christian Keimel

copy IRT

Where

What

bull Hybrid broadcast broadband technologies

bull Big Data in broadcasting environment and content-creation

bull Audio-visual QoE assessment (with crowdsourcing)

About me

Christian Keimel Slide 60

Lecturer for Digital Broadcast Engineering

Research Engineer for Media Services and Platforms

copy IRT

Crowdsourcing in large scale optimisation

The human component of Big Data

Moving crowdsourcing to lean-back devices

Crowdsourcing from the couch

Crowdsourced experiments emulating the laboratory

Is it time for new methodologies ignoring the lab

Questionshellip

Christian Keimel Slide 61

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 24: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Min Chen Professor of Scientific Visualization

University of Oxford (2011 ~ Present) Swansea University (1984 ~ 2011)

Research Interests most aspects of visualization aspects of HCI computer graphics computer vision data mining and information theory

Research Highlight

1 Volume Graphics and Visualization

2 Video Visualization

3 Visual Analytics

4 Perception and Cognition in Visualization

5 Theory of Visualization

1 2

3 4

5

Research Question in the Context of 15481

Short Version How can we use crowdsourcing to capture perceptual and cognitive measures to facilitate mathematical modelling of visualization and interaction

Long Version Information theory is perhaps the most promising mathematical framework that can underpin the fields of visualization and interaction as well as data processing and communication However to obtain its core measures various types of entropy one needs to capture a probability distribution of all states associated with a concept While such a probability distribution can be captured in a machine-centric data processing or communication pipeline it is not trivial to capture in a human-centric process Crowdsourcing offers an opportunity to conduct ldquobigrdquo observation and measurement of perceptual and cognitive activities It is thus a tool for ldquoBig Sciencerdquo

Stephan Diehl

Dagstuhl Seminar 15481 Evaluation in the Crowd Crowdsourcing and Human-Centred Experiments

Stephan Diehl Software Engineering University of Trier Germany

The Java Code Clone

Detection API

Empirical Software Engineering bull [JSS2015 ESEM2015 FSE2014 ICSM 2012

FSE 2011 ESEM 2011 hellip]

Empirical Information Visualization bull [InfoVis 2010 J of Media Psychology 410 hellip]

Analysis and Collaboration SE Tools bull [ASE2014 ASE 2012 ICSE 2011 TSE 2005 ]

Visualizing (Dynamic) Compound Graphs bull [EuroVisSTAR 2014 IV Journal 2012 InfoVis11

SOFTVIS10 EuroVis09 IV09 EuroVis08

CHASE08 AVI08GD04GD02 hellip]

Dagstuhl Seminar 15481 Evaluation in the Crowd Crowdsourcing and Human-Centred Experiments

bull How to reach your (very specific) target audience without

spamming

bull How to control for resolution animation speed color

bull What crowdsourcing platforms exist and what are their

features

Darren Edwards

Darren Edwards Swansea University

Cognitive psychology - categorization

Health and clinical psychology ndash applications of categorization on clinical populations (eg autism)

Crowdsourcing and psychological interventions

Recent validity studies confirm the use of simple psychological tasks such as decision making

Recent attempts have now explored more complex tasks such as psychological interventions (eg mindfulness)

Present studies now explore more complex forms of mindfulness

What are the limitations of using crowdsourced populations online such as when using psychological interventions or categorization

Sebastian Egger

Engineering

Technical high school

bull Telematics

electrotechnical enginieering telecommunications

bull Signal Processing for UWB systems

Sociology

Theory of human societies

Interaction amp collaboration behaviour of groups

Measurement of human behaviour

Currently Human Computer Interaction

Social amp collaborative aspects of technology use

33 26112015

Sebastian Egger

Motivational aspects of crowdworkers

CS as a collaborative work setting

Optimal Setup of CS experiments

Questions (max Nr of questions max duration)

Gamified reliability checks

Scale designs

CS vs Lab whatlsquos the ground truth

New application domains of CS

34 26112015

Interest in Crowdsourcing

Sara Fabrikant

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

humanmdashsystem (visuo-spatial) interface

spatio-temporal analytics

spatialization human navigation

multivariate spatial analysis etc)

interface design

large amp small interactive (map) displays

(ie desktop mobile VR etc)

empirical evaluations

(ie eye tracking EDA EEG etc) [httpwwwunitsevenconz]

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

human sensing observatory set up

controlled laboratory | in-situ (messy) real world setting

semi|automatic data collection of various human behavioral data streams

ie psycho-physiological perceptual cognitive and sensory-motor behaviors (through crowdsourcing )

Brian Fisher

My double research life

Brian Fisher

bull PhD in Experimental Psych

bull Postdoc w Cognitive Science

society founder amp president

bull Psychonomics Fellow

bull VIS-related symposia at Cogsci

amp APS papers on cogsci of

interaction

bull Fuzzy-logicBayes models

bull Postdoc funded by Inst for Robotics

and AI

bull VAST SC VEC VACCINE

bull Co-organized Dagstuhl ldquoInteraction

with Information for Visual Reasoningrdquo

Cogsci-based papers at VIS CHI

BELIV

My Plan

VIS

Me Psychological Science

Ujwal Gadiraju

p = 0055111

Crowd Worker

Behavior

Microtask

Design

Crowdsourcing Paradigm Quality amp Effectiveness

UJWAL GADIRAJU

Incentivization Gamification

Neha Gupta

Getting to know the crowd

Presented at Dagstuhl by

Dr David Martin amp Neha Gupta

Neha Gupta

PhD Student School of Computer Science University of Nottingham UK

Investigating crowdwork through the platform Amazon Mechanical Turk

psxng1nottinghamacuk

Neha_Gi_ji

Research amp Research teams

Research Studies

bull Qualitative study of Turkers on online forums like mturkforum and Turker Nation by Dr David Martin XRCE

bull Ethnographic Studies of Turkers in their respective homes and places of work in India by Neha Gupta UoN

Research teams

Xerox

bull David Martin Jacki OrsquoNeill (Now MSR India) Ben Hanrahan (Now PSU)

psxng1nottinghamacuk

Neha_Gi_ji

Nottingham

bull Andy Crabtree

Tom Rodden

Matthias Hirth

Matthias Hirth 48 48

About Myself

Matthias Hirth

Research Assistant at University of Wuumlrzburg

httpmatthias-hirthcom

matthiashirthinformatikuni-wuerzburgde

Crowdsourcing

Platform analysis

Platform optimization

Human factors

Use cases

Online Social Networks

Collaboration networks

Structural differences of OSNs

Application of OSN relationships

Interactive applicationsGaming

Challenges of real time interactions

Serious gamingGamification

E-Sports broadcastinglive streaming

Quality of Experience

Monitoring techniques

Quantification techniques

(Crowd-bases) test design

Social interactions

Test methodology Incentive design

Use case

Measurement technique

Identification of

qualification

Matthias Hirth 49 49

Improving Crowdsourcing Technology and the Way We Use it

Crowdsourcing technology

How to transfer well known lab procedures into an uncontrolled

crowdsourcing environment

Which crowdsourcing platform features are required to conduct

successful experiments but are not present yet

Which technical developments (can) foster the usage of

crowdsourcing instead of lab experiments

How can technological developments be used for novel incentive

mechanisms

Usage of Crowdsourcing

How can we conduct reproducible experiments in diverse

crowdsourcing environment

How can we raise the awareness of crowdsourcing employers for the

individuals behind the crowd

Tobias Hoszligfeld

The crowd community in the lab

Tobias Hoszligfeld wwwmaswiwiuni-duede

bull Background Tobias Hoszligfeld

ndash 2004-2014 Research group ldquoFuture Internet Applicationsrdquo Chair of Communication Networks Univ of Wuumlrzburg

ndash Since Sep 2014 Chair of Modeling of Adaptive Systems (head) University of Duisburg-Essen

bull Interests Application of crowdsourcing

ndash User studies on Quality of Experience (QoE) for Internet apps eg video QoE in the crowd amp in the lab methodology reliability statisticshellip [2]

ndash Mobile crowdsourcing network and context measurements eg Wifi offloading incentives privacy recommendation mechanisms etc [1]

bull Interests The crowd community

ndash Crowd = Human beings Social network wisdom of the crowd community

ndash UnderstandingUtilizing the crowd community for successful human computation For which jobs can social interaction improve the quality of work Classification

ndash How can we deploy such a system where interaction is triggered

ndash System design Task design interaction mechanisms incentives

22 Nov 2015

[1] httpdropsdagstuhldeopusvolltexte20134354

[2] httpswww3informatikuni-wuerzburgdeqoewikiqualinetcrowd

Challenges Example of Crowdsourced QoE

Tobias Hoszligfeld wwwmaswiwiuni-duede

task design

incentives

Internet

crowd

community

anonymous

remote

subjects

Internet delivery of

test data and user

response

QoE study bull testing methodology bull QoE influence factors

under investigation bull additional QoE factors bull proper subset of items bull hellip

crowdsourcing setting bull short task duration bull instructions amp training bull feedback amp questions bull hidden influence factors bull reliability of users bull hellip

heterogeneous devices

and Internet access

implementation extrinsic vs intrinsic

motivation

Processing of data

Technical aspects bull server implementation bull client application amp

compatibility bull desired test conditions bull context monitoring bull hellip

test items

lab

scientific methodologies

22 Nov 2015

Jason Jacques

Jason Jacques supervised by Per Ola Kristensson

T

Crowdsourcing Code Perception

jtj21camacuk peopledscamacukjtj21

jtjacquesgmailcom jsonjcouk

PhD Working Title

science

Crowdsourcing Interests W

hat

W

ho

Ho

w

Wh

y jtj21camacuk jsonjcouk science

Radu Jianu

Radu Jianu Assistant Professor

Florida International University (PhD Brown University lsquo12)

Interests

Applications of eye-tracking in visualization tracking what people look at rather than where they look

bull Data intensive large scale eye-tracking experiments bull Gaze-contingent visualizations bull Gaze-driven data recommendation and big data curation

Automating user studies in data visualizations (on next page)

Automating user studies

1 Software designs that can (semi)automate a broad range of quantitative evaluations by leveraging

ndash Crowdsourcing ndash Benchmark data and task taxonomies ndash Automated result interpretation

2 Moving user evaluation from the end to within the software design and implementation process as a means to choose between competitive visual designs

Christian Keimel

copy IRT

Where

What

bull Hybrid broadcast broadband technologies

bull Big Data in broadcasting environment and content-creation

bull Audio-visual QoE assessment (with crowdsourcing)

About me

Christian Keimel Slide 60

Lecturer for Digital Broadcast Engineering

Research Engineer for Media Services and Platforms

copy IRT

Crowdsourcing in large scale optimisation

The human component of Big Data

Moving crowdsourcing to lean-back devices

Crowdsourcing from the couch

Crowdsourced experiments emulating the laboratory

Is it time for new methodologies ignoring the lab

Questionshellip

Christian Keimel Slide 61

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 25: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Research Question in the Context of 15481

Short Version How can we use crowdsourcing to capture perceptual and cognitive measures to facilitate mathematical modelling of visualization and interaction

Long Version Information theory is perhaps the most promising mathematical framework that can underpin the fields of visualization and interaction as well as data processing and communication However to obtain its core measures various types of entropy one needs to capture a probability distribution of all states associated with a concept While such a probability distribution can be captured in a machine-centric data processing or communication pipeline it is not trivial to capture in a human-centric process Crowdsourcing offers an opportunity to conduct ldquobigrdquo observation and measurement of perceptual and cognitive activities It is thus a tool for ldquoBig Sciencerdquo

Stephan Diehl

Dagstuhl Seminar 15481 Evaluation in the Crowd Crowdsourcing and Human-Centred Experiments

Stephan Diehl Software Engineering University of Trier Germany

The Java Code Clone

Detection API

Empirical Software Engineering bull [JSS2015 ESEM2015 FSE2014 ICSM 2012

FSE 2011 ESEM 2011 hellip]

Empirical Information Visualization bull [InfoVis 2010 J of Media Psychology 410 hellip]

Analysis and Collaboration SE Tools bull [ASE2014 ASE 2012 ICSE 2011 TSE 2005 ]

Visualizing (Dynamic) Compound Graphs bull [EuroVisSTAR 2014 IV Journal 2012 InfoVis11

SOFTVIS10 EuroVis09 IV09 EuroVis08

CHASE08 AVI08GD04GD02 hellip]

Dagstuhl Seminar 15481 Evaluation in the Crowd Crowdsourcing and Human-Centred Experiments

bull How to reach your (very specific) target audience without

spamming

bull How to control for resolution animation speed color

bull What crowdsourcing platforms exist and what are their

features

Darren Edwards

Darren Edwards Swansea University

Cognitive psychology - categorization

Health and clinical psychology ndash applications of categorization on clinical populations (eg autism)

Crowdsourcing and psychological interventions

Recent validity studies confirm the use of simple psychological tasks such as decision making

Recent attempts have now explored more complex tasks such as psychological interventions (eg mindfulness)

Present studies now explore more complex forms of mindfulness

What are the limitations of using crowdsourced populations online such as when using psychological interventions or categorization

Sebastian Egger

Engineering

Technical high school

bull Telematics

electrotechnical enginieering telecommunications

bull Signal Processing for UWB systems

Sociology

Theory of human societies

Interaction amp collaboration behaviour of groups

Measurement of human behaviour

Currently Human Computer Interaction

Social amp collaborative aspects of technology use

33 26112015

Sebastian Egger

Motivational aspects of crowdworkers

CS as a collaborative work setting

Optimal Setup of CS experiments

Questions (max Nr of questions max duration)

Gamified reliability checks

Scale designs

CS vs Lab whatlsquos the ground truth

New application domains of CS

34 26112015

Interest in Crowdsourcing

Sara Fabrikant

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

humanmdashsystem (visuo-spatial) interface

spatio-temporal analytics

spatialization human navigation

multivariate spatial analysis etc)

interface design

large amp small interactive (map) displays

(ie desktop mobile VR etc)

empirical evaluations

(ie eye tracking EDA EEG etc) [httpwwwunitsevenconz]

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

human sensing observatory set up

controlled laboratory | in-situ (messy) real world setting

semi|automatic data collection of various human behavioral data streams

ie psycho-physiological perceptual cognitive and sensory-motor behaviors (through crowdsourcing )

Brian Fisher

My double research life

Brian Fisher

bull PhD in Experimental Psych

bull Postdoc w Cognitive Science

society founder amp president

bull Psychonomics Fellow

bull VIS-related symposia at Cogsci

amp APS papers on cogsci of

interaction

bull Fuzzy-logicBayes models

bull Postdoc funded by Inst for Robotics

and AI

bull VAST SC VEC VACCINE

bull Co-organized Dagstuhl ldquoInteraction

with Information for Visual Reasoningrdquo

Cogsci-based papers at VIS CHI

BELIV

My Plan

VIS

Me Psychological Science

Ujwal Gadiraju

p = 0055111

Crowd Worker

Behavior

Microtask

Design

Crowdsourcing Paradigm Quality amp Effectiveness

UJWAL GADIRAJU

Incentivization Gamification

Neha Gupta

Getting to know the crowd

Presented at Dagstuhl by

Dr David Martin amp Neha Gupta

Neha Gupta

PhD Student School of Computer Science University of Nottingham UK

Investigating crowdwork through the platform Amazon Mechanical Turk

psxng1nottinghamacuk

Neha_Gi_ji

Research amp Research teams

Research Studies

bull Qualitative study of Turkers on online forums like mturkforum and Turker Nation by Dr David Martin XRCE

bull Ethnographic Studies of Turkers in their respective homes and places of work in India by Neha Gupta UoN

Research teams

Xerox

bull David Martin Jacki OrsquoNeill (Now MSR India) Ben Hanrahan (Now PSU)

psxng1nottinghamacuk

Neha_Gi_ji

Nottingham

bull Andy Crabtree

Tom Rodden

Matthias Hirth

Matthias Hirth 48 48

About Myself

Matthias Hirth

Research Assistant at University of Wuumlrzburg

httpmatthias-hirthcom

matthiashirthinformatikuni-wuerzburgde

Crowdsourcing

Platform analysis

Platform optimization

Human factors

Use cases

Online Social Networks

Collaboration networks

Structural differences of OSNs

Application of OSN relationships

Interactive applicationsGaming

Challenges of real time interactions

Serious gamingGamification

E-Sports broadcastinglive streaming

Quality of Experience

Monitoring techniques

Quantification techniques

(Crowd-bases) test design

Social interactions

Test methodology Incentive design

Use case

Measurement technique

Identification of

qualification

Matthias Hirth 49 49

Improving Crowdsourcing Technology and the Way We Use it

Crowdsourcing technology

How to transfer well known lab procedures into an uncontrolled

crowdsourcing environment

Which crowdsourcing platform features are required to conduct

successful experiments but are not present yet

Which technical developments (can) foster the usage of

crowdsourcing instead of lab experiments

How can technological developments be used for novel incentive

mechanisms

Usage of Crowdsourcing

How can we conduct reproducible experiments in diverse

crowdsourcing environment

How can we raise the awareness of crowdsourcing employers for the

individuals behind the crowd

Tobias Hoszligfeld

The crowd community in the lab

Tobias Hoszligfeld wwwmaswiwiuni-duede

bull Background Tobias Hoszligfeld

ndash 2004-2014 Research group ldquoFuture Internet Applicationsrdquo Chair of Communication Networks Univ of Wuumlrzburg

ndash Since Sep 2014 Chair of Modeling of Adaptive Systems (head) University of Duisburg-Essen

bull Interests Application of crowdsourcing

ndash User studies on Quality of Experience (QoE) for Internet apps eg video QoE in the crowd amp in the lab methodology reliability statisticshellip [2]

ndash Mobile crowdsourcing network and context measurements eg Wifi offloading incentives privacy recommendation mechanisms etc [1]

bull Interests The crowd community

ndash Crowd = Human beings Social network wisdom of the crowd community

ndash UnderstandingUtilizing the crowd community for successful human computation For which jobs can social interaction improve the quality of work Classification

ndash How can we deploy such a system where interaction is triggered

ndash System design Task design interaction mechanisms incentives

22 Nov 2015

[1] httpdropsdagstuhldeopusvolltexte20134354

[2] httpswww3informatikuni-wuerzburgdeqoewikiqualinetcrowd

Challenges Example of Crowdsourced QoE

Tobias Hoszligfeld wwwmaswiwiuni-duede

task design

incentives

Internet

crowd

community

anonymous

remote

subjects

Internet delivery of

test data and user

response

QoE study bull testing methodology bull QoE influence factors

under investigation bull additional QoE factors bull proper subset of items bull hellip

crowdsourcing setting bull short task duration bull instructions amp training bull feedback amp questions bull hidden influence factors bull reliability of users bull hellip

heterogeneous devices

and Internet access

implementation extrinsic vs intrinsic

motivation

Processing of data

Technical aspects bull server implementation bull client application amp

compatibility bull desired test conditions bull context monitoring bull hellip

test items

lab

scientific methodologies

22 Nov 2015

Jason Jacques

Jason Jacques supervised by Per Ola Kristensson

T

Crowdsourcing Code Perception

jtj21camacuk peopledscamacukjtj21

jtjacquesgmailcom jsonjcouk

PhD Working Title

science

Crowdsourcing Interests W

hat

W

ho

Ho

w

Wh

y jtj21camacuk jsonjcouk science

Radu Jianu

Radu Jianu Assistant Professor

Florida International University (PhD Brown University lsquo12)

Interests

Applications of eye-tracking in visualization tracking what people look at rather than where they look

bull Data intensive large scale eye-tracking experiments bull Gaze-contingent visualizations bull Gaze-driven data recommendation and big data curation

Automating user studies in data visualizations (on next page)

Automating user studies

1 Software designs that can (semi)automate a broad range of quantitative evaluations by leveraging

ndash Crowdsourcing ndash Benchmark data and task taxonomies ndash Automated result interpretation

2 Moving user evaluation from the end to within the software design and implementation process as a means to choose between competitive visual designs

Christian Keimel

copy IRT

Where

What

bull Hybrid broadcast broadband technologies

bull Big Data in broadcasting environment and content-creation

bull Audio-visual QoE assessment (with crowdsourcing)

About me

Christian Keimel Slide 60

Lecturer for Digital Broadcast Engineering

Research Engineer for Media Services and Platforms

copy IRT

Crowdsourcing in large scale optimisation

The human component of Big Data

Moving crowdsourcing to lean-back devices

Crowdsourcing from the couch

Crowdsourced experiments emulating the laboratory

Is it time for new methodologies ignoring the lab

Questionshellip

Christian Keimel Slide 61

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 26: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Stephan Diehl

Dagstuhl Seminar 15481 Evaluation in the Crowd Crowdsourcing and Human-Centred Experiments

Stephan Diehl Software Engineering University of Trier Germany

The Java Code Clone

Detection API

Empirical Software Engineering bull [JSS2015 ESEM2015 FSE2014 ICSM 2012

FSE 2011 ESEM 2011 hellip]

Empirical Information Visualization bull [InfoVis 2010 J of Media Psychology 410 hellip]

Analysis and Collaboration SE Tools bull [ASE2014 ASE 2012 ICSE 2011 TSE 2005 ]

Visualizing (Dynamic) Compound Graphs bull [EuroVisSTAR 2014 IV Journal 2012 InfoVis11

SOFTVIS10 EuroVis09 IV09 EuroVis08

CHASE08 AVI08GD04GD02 hellip]

Dagstuhl Seminar 15481 Evaluation in the Crowd Crowdsourcing and Human-Centred Experiments

bull How to reach your (very specific) target audience without

spamming

bull How to control for resolution animation speed color

bull What crowdsourcing platforms exist and what are their

features

Darren Edwards

Darren Edwards Swansea University

Cognitive psychology - categorization

Health and clinical psychology ndash applications of categorization on clinical populations (eg autism)

Crowdsourcing and psychological interventions

Recent validity studies confirm the use of simple psychological tasks such as decision making

Recent attempts have now explored more complex tasks such as psychological interventions (eg mindfulness)

Present studies now explore more complex forms of mindfulness

What are the limitations of using crowdsourced populations online such as when using psychological interventions or categorization

Sebastian Egger

Engineering

Technical high school

bull Telematics

electrotechnical enginieering telecommunications

bull Signal Processing for UWB systems

Sociology

Theory of human societies

Interaction amp collaboration behaviour of groups

Measurement of human behaviour

Currently Human Computer Interaction

Social amp collaborative aspects of technology use

33 26112015

Sebastian Egger

Motivational aspects of crowdworkers

CS as a collaborative work setting

Optimal Setup of CS experiments

Questions (max Nr of questions max duration)

Gamified reliability checks

Scale designs

CS vs Lab whatlsquos the ground truth

New application domains of CS

34 26112015

Interest in Crowdsourcing

Sara Fabrikant

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

humanmdashsystem (visuo-spatial) interface

spatio-temporal analytics

spatialization human navigation

multivariate spatial analysis etc)

interface design

large amp small interactive (map) displays

(ie desktop mobile VR etc)

empirical evaluations

(ie eye tracking EDA EEG etc) [httpwwwunitsevenconz]

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

human sensing observatory set up

controlled laboratory | in-situ (messy) real world setting

semi|automatic data collection of various human behavioral data streams

ie psycho-physiological perceptual cognitive and sensory-motor behaviors (through crowdsourcing )

Brian Fisher

My double research life

Brian Fisher

bull PhD in Experimental Psych

bull Postdoc w Cognitive Science

society founder amp president

bull Psychonomics Fellow

bull VIS-related symposia at Cogsci

amp APS papers on cogsci of

interaction

bull Fuzzy-logicBayes models

bull Postdoc funded by Inst for Robotics

and AI

bull VAST SC VEC VACCINE

bull Co-organized Dagstuhl ldquoInteraction

with Information for Visual Reasoningrdquo

Cogsci-based papers at VIS CHI

BELIV

My Plan

VIS

Me Psychological Science

Ujwal Gadiraju

p = 0055111

Crowd Worker

Behavior

Microtask

Design

Crowdsourcing Paradigm Quality amp Effectiveness

UJWAL GADIRAJU

Incentivization Gamification

Neha Gupta

Getting to know the crowd

Presented at Dagstuhl by

Dr David Martin amp Neha Gupta

Neha Gupta

PhD Student School of Computer Science University of Nottingham UK

Investigating crowdwork through the platform Amazon Mechanical Turk

psxng1nottinghamacuk

Neha_Gi_ji

Research amp Research teams

Research Studies

bull Qualitative study of Turkers on online forums like mturkforum and Turker Nation by Dr David Martin XRCE

bull Ethnographic Studies of Turkers in their respective homes and places of work in India by Neha Gupta UoN

Research teams

Xerox

bull David Martin Jacki OrsquoNeill (Now MSR India) Ben Hanrahan (Now PSU)

psxng1nottinghamacuk

Neha_Gi_ji

Nottingham

bull Andy Crabtree

Tom Rodden

Matthias Hirth

Matthias Hirth 48 48

About Myself

Matthias Hirth

Research Assistant at University of Wuumlrzburg

httpmatthias-hirthcom

matthiashirthinformatikuni-wuerzburgde

Crowdsourcing

Platform analysis

Platform optimization

Human factors

Use cases

Online Social Networks

Collaboration networks

Structural differences of OSNs

Application of OSN relationships

Interactive applicationsGaming

Challenges of real time interactions

Serious gamingGamification

E-Sports broadcastinglive streaming

Quality of Experience

Monitoring techniques

Quantification techniques

(Crowd-bases) test design

Social interactions

Test methodology Incentive design

Use case

Measurement technique

Identification of

qualification

Matthias Hirth 49 49

Improving Crowdsourcing Technology and the Way We Use it

Crowdsourcing technology

How to transfer well known lab procedures into an uncontrolled

crowdsourcing environment

Which crowdsourcing platform features are required to conduct

successful experiments but are not present yet

Which technical developments (can) foster the usage of

crowdsourcing instead of lab experiments

How can technological developments be used for novel incentive

mechanisms

Usage of Crowdsourcing

How can we conduct reproducible experiments in diverse

crowdsourcing environment

How can we raise the awareness of crowdsourcing employers for the

individuals behind the crowd

Tobias Hoszligfeld

The crowd community in the lab

Tobias Hoszligfeld wwwmaswiwiuni-duede

bull Background Tobias Hoszligfeld

ndash 2004-2014 Research group ldquoFuture Internet Applicationsrdquo Chair of Communication Networks Univ of Wuumlrzburg

ndash Since Sep 2014 Chair of Modeling of Adaptive Systems (head) University of Duisburg-Essen

bull Interests Application of crowdsourcing

ndash User studies on Quality of Experience (QoE) for Internet apps eg video QoE in the crowd amp in the lab methodology reliability statisticshellip [2]

ndash Mobile crowdsourcing network and context measurements eg Wifi offloading incentives privacy recommendation mechanisms etc [1]

bull Interests The crowd community

ndash Crowd = Human beings Social network wisdom of the crowd community

ndash UnderstandingUtilizing the crowd community for successful human computation For which jobs can social interaction improve the quality of work Classification

ndash How can we deploy such a system where interaction is triggered

ndash System design Task design interaction mechanisms incentives

22 Nov 2015

[1] httpdropsdagstuhldeopusvolltexte20134354

[2] httpswww3informatikuni-wuerzburgdeqoewikiqualinetcrowd

Challenges Example of Crowdsourced QoE

Tobias Hoszligfeld wwwmaswiwiuni-duede

task design

incentives

Internet

crowd

community

anonymous

remote

subjects

Internet delivery of

test data and user

response

QoE study bull testing methodology bull QoE influence factors

under investigation bull additional QoE factors bull proper subset of items bull hellip

crowdsourcing setting bull short task duration bull instructions amp training bull feedback amp questions bull hidden influence factors bull reliability of users bull hellip

heterogeneous devices

and Internet access

implementation extrinsic vs intrinsic

motivation

Processing of data

Technical aspects bull server implementation bull client application amp

compatibility bull desired test conditions bull context monitoring bull hellip

test items

lab

scientific methodologies

22 Nov 2015

Jason Jacques

Jason Jacques supervised by Per Ola Kristensson

T

Crowdsourcing Code Perception

jtj21camacuk peopledscamacukjtj21

jtjacquesgmailcom jsonjcouk

PhD Working Title

science

Crowdsourcing Interests W

hat

W

ho

Ho

w

Wh

y jtj21camacuk jsonjcouk science

Radu Jianu

Radu Jianu Assistant Professor

Florida International University (PhD Brown University lsquo12)

Interests

Applications of eye-tracking in visualization tracking what people look at rather than where they look

bull Data intensive large scale eye-tracking experiments bull Gaze-contingent visualizations bull Gaze-driven data recommendation and big data curation

Automating user studies in data visualizations (on next page)

Automating user studies

1 Software designs that can (semi)automate a broad range of quantitative evaluations by leveraging

ndash Crowdsourcing ndash Benchmark data and task taxonomies ndash Automated result interpretation

2 Moving user evaluation from the end to within the software design and implementation process as a means to choose between competitive visual designs

Christian Keimel

copy IRT

Where

What

bull Hybrid broadcast broadband technologies

bull Big Data in broadcasting environment and content-creation

bull Audio-visual QoE assessment (with crowdsourcing)

About me

Christian Keimel Slide 60

Lecturer for Digital Broadcast Engineering

Research Engineer for Media Services and Platforms

copy IRT

Crowdsourcing in large scale optimisation

The human component of Big Data

Moving crowdsourcing to lean-back devices

Crowdsourcing from the couch

Crowdsourced experiments emulating the laboratory

Is it time for new methodologies ignoring the lab

Questionshellip

Christian Keimel Slide 61

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 27: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Dagstuhl Seminar 15481 Evaluation in the Crowd Crowdsourcing and Human-Centred Experiments

Stephan Diehl Software Engineering University of Trier Germany

The Java Code Clone

Detection API

Empirical Software Engineering bull [JSS2015 ESEM2015 FSE2014 ICSM 2012

FSE 2011 ESEM 2011 hellip]

Empirical Information Visualization bull [InfoVis 2010 J of Media Psychology 410 hellip]

Analysis and Collaboration SE Tools bull [ASE2014 ASE 2012 ICSE 2011 TSE 2005 ]

Visualizing (Dynamic) Compound Graphs bull [EuroVisSTAR 2014 IV Journal 2012 InfoVis11

SOFTVIS10 EuroVis09 IV09 EuroVis08

CHASE08 AVI08GD04GD02 hellip]

Dagstuhl Seminar 15481 Evaluation in the Crowd Crowdsourcing and Human-Centred Experiments

bull How to reach your (very specific) target audience without

spamming

bull How to control for resolution animation speed color

bull What crowdsourcing platforms exist and what are their

features

Darren Edwards

Darren Edwards Swansea University

Cognitive psychology - categorization

Health and clinical psychology ndash applications of categorization on clinical populations (eg autism)

Crowdsourcing and psychological interventions

Recent validity studies confirm the use of simple psychological tasks such as decision making

Recent attempts have now explored more complex tasks such as psychological interventions (eg mindfulness)

Present studies now explore more complex forms of mindfulness

What are the limitations of using crowdsourced populations online such as when using psychological interventions or categorization

Sebastian Egger

Engineering

Technical high school

bull Telematics

electrotechnical enginieering telecommunications

bull Signal Processing for UWB systems

Sociology

Theory of human societies

Interaction amp collaboration behaviour of groups

Measurement of human behaviour

Currently Human Computer Interaction

Social amp collaborative aspects of technology use

33 26112015

Sebastian Egger

Motivational aspects of crowdworkers

CS as a collaborative work setting

Optimal Setup of CS experiments

Questions (max Nr of questions max duration)

Gamified reliability checks

Scale designs

CS vs Lab whatlsquos the ground truth

New application domains of CS

34 26112015

Interest in Crowdsourcing

Sara Fabrikant

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

humanmdashsystem (visuo-spatial) interface

spatio-temporal analytics

spatialization human navigation

multivariate spatial analysis etc)

interface design

large amp small interactive (map) displays

(ie desktop mobile VR etc)

empirical evaluations

(ie eye tracking EDA EEG etc) [httpwwwunitsevenconz]

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

human sensing observatory set up

controlled laboratory | in-situ (messy) real world setting

semi|automatic data collection of various human behavioral data streams

ie psycho-physiological perceptual cognitive and sensory-motor behaviors (through crowdsourcing )

Brian Fisher

My double research life

Brian Fisher

bull PhD in Experimental Psych

bull Postdoc w Cognitive Science

society founder amp president

bull Psychonomics Fellow

bull VIS-related symposia at Cogsci

amp APS papers on cogsci of

interaction

bull Fuzzy-logicBayes models

bull Postdoc funded by Inst for Robotics

and AI

bull VAST SC VEC VACCINE

bull Co-organized Dagstuhl ldquoInteraction

with Information for Visual Reasoningrdquo

Cogsci-based papers at VIS CHI

BELIV

My Plan

VIS

Me Psychological Science

Ujwal Gadiraju

p = 0055111

Crowd Worker

Behavior

Microtask

Design

Crowdsourcing Paradigm Quality amp Effectiveness

UJWAL GADIRAJU

Incentivization Gamification

Neha Gupta

Getting to know the crowd

Presented at Dagstuhl by

Dr David Martin amp Neha Gupta

Neha Gupta

PhD Student School of Computer Science University of Nottingham UK

Investigating crowdwork through the platform Amazon Mechanical Turk

psxng1nottinghamacuk

Neha_Gi_ji

Research amp Research teams

Research Studies

bull Qualitative study of Turkers on online forums like mturkforum and Turker Nation by Dr David Martin XRCE

bull Ethnographic Studies of Turkers in their respective homes and places of work in India by Neha Gupta UoN

Research teams

Xerox

bull David Martin Jacki OrsquoNeill (Now MSR India) Ben Hanrahan (Now PSU)

psxng1nottinghamacuk

Neha_Gi_ji

Nottingham

bull Andy Crabtree

Tom Rodden

Matthias Hirth

Matthias Hirth 48 48

About Myself

Matthias Hirth

Research Assistant at University of Wuumlrzburg

httpmatthias-hirthcom

matthiashirthinformatikuni-wuerzburgde

Crowdsourcing

Platform analysis

Platform optimization

Human factors

Use cases

Online Social Networks

Collaboration networks

Structural differences of OSNs

Application of OSN relationships

Interactive applicationsGaming

Challenges of real time interactions

Serious gamingGamification

E-Sports broadcastinglive streaming

Quality of Experience

Monitoring techniques

Quantification techniques

(Crowd-bases) test design

Social interactions

Test methodology Incentive design

Use case

Measurement technique

Identification of

qualification

Matthias Hirth 49 49

Improving Crowdsourcing Technology and the Way We Use it

Crowdsourcing technology

How to transfer well known lab procedures into an uncontrolled

crowdsourcing environment

Which crowdsourcing platform features are required to conduct

successful experiments but are not present yet

Which technical developments (can) foster the usage of

crowdsourcing instead of lab experiments

How can technological developments be used for novel incentive

mechanisms

Usage of Crowdsourcing

How can we conduct reproducible experiments in diverse

crowdsourcing environment

How can we raise the awareness of crowdsourcing employers for the

individuals behind the crowd

Tobias Hoszligfeld

The crowd community in the lab

Tobias Hoszligfeld wwwmaswiwiuni-duede

bull Background Tobias Hoszligfeld

ndash 2004-2014 Research group ldquoFuture Internet Applicationsrdquo Chair of Communication Networks Univ of Wuumlrzburg

ndash Since Sep 2014 Chair of Modeling of Adaptive Systems (head) University of Duisburg-Essen

bull Interests Application of crowdsourcing

ndash User studies on Quality of Experience (QoE) for Internet apps eg video QoE in the crowd amp in the lab methodology reliability statisticshellip [2]

ndash Mobile crowdsourcing network and context measurements eg Wifi offloading incentives privacy recommendation mechanisms etc [1]

bull Interests The crowd community

ndash Crowd = Human beings Social network wisdom of the crowd community

ndash UnderstandingUtilizing the crowd community for successful human computation For which jobs can social interaction improve the quality of work Classification

ndash How can we deploy such a system where interaction is triggered

ndash System design Task design interaction mechanisms incentives

22 Nov 2015

[1] httpdropsdagstuhldeopusvolltexte20134354

[2] httpswww3informatikuni-wuerzburgdeqoewikiqualinetcrowd

Challenges Example of Crowdsourced QoE

Tobias Hoszligfeld wwwmaswiwiuni-duede

task design

incentives

Internet

crowd

community

anonymous

remote

subjects

Internet delivery of

test data and user

response

QoE study bull testing methodology bull QoE influence factors

under investigation bull additional QoE factors bull proper subset of items bull hellip

crowdsourcing setting bull short task duration bull instructions amp training bull feedback amp questions bull hidden influence factors bull reliability of users bull hellip

heterogeneous devices

and Internet access

implementation extrinsic vs intrinsic

motivation

Processing of data

Technical aspects bull server implementation bull client application amp

compatibility bull desired test conditions bull context monitoring bull hellip

test items

lab

scientific methodologies

22 Nov 2015

Jason Jacques

Jason Jacques supervised by Per Ola Kristensson

T

Crowdsourcing Code Perception

jtj21camacuk peopledscamacukjtj21

jtjacquesgmailcom jsonjcouk

PhD Working Title

science

Crowdsourcing Interests W

hat

W

ho

Ho

w

Wh

y jtj21camacuk jsonjcouk science

Radu Jianu

Radu Jianu Assistant Professor

Florida International University (PhD Brown University lsquo12)

Interests

Applications of eye-tracking in visualization tracking what people look at rather than where they look

bull Data intensive large scale eye-tracking experiments bull Gaze-contingent visualizations bull Gaze-driven data recommendation and big data curation

Automating user studies in data visualizations (on next page)

Automating user studies

1 Software designs that can (semi)automate a broad range of quantitative evaluations by leveraging

ndash Crowdsourcing ndash Benchmark data and task taxonomies ndash Automated result interpretation

2 Moving user evaluation from the end to within the software design and implementation process as a means to choose between competitive visual designs

Christian Keimel

copy IRT

Where

What

bull Hybrid broadcast broadband technologies

bull Big Data in broadcasting environment and content-creation

bull Audio-visual QoE assessment (with crowdsourcing)

About me

Christian Keimel Slide 60

Lecturer for Digital Broadcast Engineering

Research Engineer for Media Services and Platforms

copy IRT

Crowdsourcing in large scale optimisation

The human component of Big Data

Moving crowdsourcing to lean-back devices

Crowdsourcing from the couch

Crowdsourced experiments emulating the laboratory

Is it time for new methodologies ignoring the lab

Questionshellip

Christian Keimel Slide 61

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 28: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Dagstuhl Seminar 15481 Evaluation in the Crowd Crowdsourcing and Human-Centred Experiments

bull How to reach your (very specific) target audience without

spamming

bull How to control for resolution animation speed color

bull What crowdsourcing platforms exist and what are their

features

Darren Edwards

Darren Edwards Swansea University

Cognitive psychology - categorization

Health and clinical psychology ndash applications of categorization on clinical populations (eg autism)

Crowdsourcing and psychological interventions

Recent validity studies confirm the use of simple psychological tasks such as decision making

Recent attempts have now explored more complex tasks such as psychological interventions (eg mindfulness)

Present studies now explore more complex forms of mindfulness

What are the limitations of using crowdsourced populations online such as when using psychological interventions or categorization

Sebastian Egger

Engineering

Technical high school

bull Telematics

electrotechnical enginieering telecommunications

bull Signal Processing for UWB systems

Sociology

Theory of human societies

Interaction amp collaboration behaviour of groups

Measurement of human behaviour

Currently Human Computer Interaction

Social amp collaborative aspects of technology use

33 26112015

Sebastian Egger

Motivational aspects of crowdworkers

CS as a collaborative work setting

Optimal Setup of CS experiments

Questions (max Nr of questions max duration)

Gamified reliability checks

Scale designs

CS vs Lab whatlsquos the ground truth

New application domains of CS

34 26112015

Interest in Crowdsourcing

Sara Fabrikant

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

humanmdashsystem (visuo-spatial) interface

spatio-temporal analytics

spatialization human navigation

multivariate spatial analysis etc)

interface design

large amp small interactive (map) displays

(ie desktop mobile VR etc)

empirical evaluations

(ie eye tracking EDA EEG etc) [httpwwwunitsevenconz]

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

human sensing observatory set up

controlled laboratory | in-situ (messy) real world setting

semi|automatic data collection of various human behavioral data streams

ie psycho-physiological perceptual cognitive and sensory-motor behaviors (through crowdsourcing )

Brian Fisher

My double research life

Brian Fisher

bull PhD in Experimental Psych

bull Postdoc w Cognitive Science

society founder amp president

bull Psychonomics Fellow

bull VIS-related symposia at Cogsci

amp APS papers on cogsci of

interaction

bull Fuzzy-logicBayes models

bull Postdoc funded by Inst for Robotics

and AI

bull VAST SC VEC VACCINE

bull Co-organized Dagstuhl ldquoInteraction

with Information for Visual Reasoningrdquo

Cogsci-based papers at VIS CHI

BELIV

My Plan

VIS

Me Psychological Science

Ujwal Gadiraju

p = 0055111

Crowd Worker

Behavior

Microtask

Design

Crowdsourcing Paradigm Quality amp Effectiveness

UJWAL GADIRAJU

Incentivization Gamification

Neha Gupta

Getting to know the crowd

Presented at Dagstuhl by

Dr David Martin amp Neha Gupta

Neha Gupta

PhD Student School of Computer Science University of Nottingham UK

Investigating crowdwork through the platform Amazon Mechanical Turk

psxng1nottinghamacuk

Neha_Gi_ji

Research amp Research teams

Research Studies

bull Qualitative study of Turkers on online forums like mturkforum and Turker Nation by Dr David Martin XRCE

bull Ethnographic Studies of Turkers in their respective homes and places of work in India by Neha Gupta UoN

Research teams

Xerox

bull David Martin Jacki OrsquoNeill (Now MSR India) Ben Hanrahan (Now PSU)

psxng1nottinghamacuk

Neha_Gi_ji

Nottingham

bull Andy Crabtree

Tom Rodden

Matthias Hirth

Matthias Hirth 48 48

About Myself

Matthias Hirth

Research Assistant at University of Wuumlrzburg

httpmatthias-hirthcom

matthiashirthinformatikuni-wuerzburgde

Crowdsourcing

Platform analysis

Platform optimization

Human factors

Use cases

Online Social Networks

Collaboration networks

Structural differences of OSNs

Application of OSN relationships

Interactive applicationsGaming

Challenges of real time interactions

Serious gamingGamification

E-Sports broadcastinglive streaming

Quality of Experience

Monitoring techniques

Quantification techniques

(Crowd-bases) test design

Social interactions

Test methodology Incentive design

Use case

Measurement technique

Identification of

qualification

Matthias Hirth 49 49

Improving Crowdsourcing Technology and the Way We Use it

Crowdsourcing technology

How to transfer well known lab procedures into an uncontrolled

crowdsourcing environment

Which crowdsourcing platform features are required to conduct

successful experiments but are not present yet

Which technical developments (can) foster the usage of

crowdsourcing instead of lab experiments

How can technological developments be used for novel incentive

mechanisms

Usage of Crowdsourcing

How can we conduct reproducible experiments in diverse

crowdsourcing environment

How can we raise the awareness of crowdsourcing employers for the

individuals behind the crowd

Tobias Hoszligfeld

The crowd community in the lab

Tobias Hoszligfeld wwwmaswiwiuni-duede

bull Background Tobias Hoszligfeld

ndash 2004-2014 Research group ldquoFuture Internet Applicationsrdquo Chair of Communication Networks Univ of Wuumlrzburg

ndash Since Sep 2014 Chair of Modeling of Adaptive Systems (head) University of Duisburg-Essen

bull Interests Application of crowdsourcing

ndash User studies on Quality of Experience (QoE) for Internet apps eg video QoE in the crowd amp in the lab methodology reliability statisticshellip [2]

ndash Mobile crowdsourcing network and context measurements eg Wifi offloading incentives privacy recommendation mechanisms etc [1]

bull Interests The crowd community

ndash Crowd = Human beings Social network wisdom of the crowd community

ndash UnderstandingUtilizing the crowd community for successful human computation For which jobs can social interaction improve the quality of work Classification

ndash How can we deploy such a system where interaction is triggered

ndash System design Task design interaction mechanisms incentives

22 Nov 2015

[1] httpdropsdagstuhldeopusvolltexte20134354

[2] httpswww3informatikuni-wuerzburgdeqoewikiqualinetcrowd

Challenges Example of Crowdsourced QoE

Tobias Hoszligfeld wwwmaswiwiuni-duede

task design

incentives

Internet

crowd

community

anonymous

remote

subjects

Internet delivery of

test data and user

response

QoE study bull testing methodology bull QoE influence factors

under investigation bull additional QoE factors bull proper subset of items bull hellip

crowdsourcing setting bull short task duration bull instructions amp training bull feedback amp questions bull hidden influence factors bull reliability of users bull hellip

heterogeneous devices

and Internet access

implementation extrinsic vs intrinsic

motivation

Processing of data

Technical aspects bull server implementation bull client application amp

compatibility bull desired test conditions bull context monitoring bull hellip

test items

lab

scientific methodologies

22 Nov 2015

Jason Jacques

Jason Jacques supervised by Per Ola Kristensson

T

Crowdsourcing Code Perception

jtj21camacuk peopledscamacukjtj21

jtjacquesgmailcom jsonjcouk

PhD Working Title

science

Crowdsourcing Interests W

hat

W

ho

Ho

w

Wh

y jtj21camacuk jsonjcouk science

Radu Jianu

Radu Jianu Assistant Professor

Florida International University (PhD Brown University lsquo12)

Interests

Applications of eye-tracking in visualization tracking what people look at rather than where they look

bull Data intensive large scale eye-tracking experiments bull Gaze-contingent visualizations bull Gaze-driven data recommendation and big data curation

Automating user studies in data visualizations (on next page)

Automating user studies

1 Software designs that can (semi)automate a broad range of quantitative evaluations by leveraging

ndash Crowdsourcing ndash Benchmark data and task taxonomies ndash Automated result interpretation

2 Moving user evaluation from the end to within the software design and implementation process as a means to choose between competitive visual designs

Christian Keimel

copy IRT

Where

What

bull Hybrid broadcast broadband technologies

bull Big Data in broadcasting environment and content-creation

bull Audio-visual QoE assessment (with crowdsourcing)

About me

Christian Keimel Slide 60

Lecturer for Digital Broadcast Engineering

Research Engineer for Media Services and Platforms

copy IRT

Crowdsourcing in large scale optimisation

The human component of Big Data

Moving crowdsourcing to lean-back devices

Crowdsourcing from the couch

Crowdsourced experiments emulating the laboratory

Is it time for new methodologies ignoring the lab

Questionshellip

Christian Keimel Slide 61

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 29: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Darren Edwards

Darren Edwards Swansea University

Cognitive psychology - categorization

Health and clinical psychology ndash applications of categorization on clinical populations (eg autism)

Crowdsourcing and psychological interventions

Recent validity studies confirm the use of simple psychological tasks such as decision making

Recent attempts have now explored more complex tasks such as psychological interventions (eg mindfulness)

Present studies now explore more complex forms of mindfulness

What are the limitations of using crowdsourced populations online such as when using psychological interventions or categorization

Sebastian Egger

Engineering

Technical high school

bull Telematics

electrotechnical enginieering telecommunications

bull Signal Processing for UWB systems

Sociology

Theory of human societies

Interaction amp collaboration behaviour of groups

Measurement of human behaviour

Currently Human Computer Interaction

Social amp collaborative aspects of technology use

33 26112015

Sebastian Egger

Motivational aspects of crowdworkers

CS as a collaborative work setting

Optimal Setup of CS experiments

Questions (max Nr of questions max duration)

Gamified reliability checks

Scale designs

CS vs Lab whatlsquos the ground truth

New application domains of CS

34 26112015

Interest in Crowdsourcing

Sara Fabrikant

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

humanmdashsystem (visuo-spatial) interface

spatio-temporal analytics

spatialization human navigation

multivariate spatial analysis etc)

interface design

large amp small interactive (map) displays

(ie desktop mobile VR etc)

empirical evaluations

(ie eye tracking EDA EEG etc) [httpwwwunitsevenconz]

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

human sensing observatory set up

controlled laboratory | in-situ (messy) real world setting

semi|automatic data collection of various human behavioral data streams

ie psycho-physiological perceptual cognitive and sensory-motor behaviors (through crowdsourcing )

Brian Fisher

My double research life

Brian Fisher

bull PhD in Experimental Psych

bull Postdoc w Cognitive Science

society founder amp president

bull Psychonomics Fellow

bull VIS-related symposia at Cogsci

amp APS papers on cogsci of

interaction

bull Fuzzy-logicBayes models

bull Postdoc funded by Inst for Robotics

and AI

bull VAST SC VEC VACCINE

bull Co-organized Dagstuhl ldquoInteraction

with Information for Visual Reasoningrdquo

Cogsci-based papers at VIS CHI

BELIV

My Plan

VIS

Me Psychological Science

Ujwal Gadiraju

p = 0055111

Crowd Worker

Behavior

Microtask

Design

Crowdsourcing Paradigm Quality amp Effectiveness

UJWAL GADIRAJU

Incentivization Gamification

Neha Gupta

Getting to know the crowd

Presented at Dagstuhl by

Dr David Martin amp Neha Gupta

Neha Gupta

PhD Student School of Computer Science University of Nottingham UK

Investigating crowdwork through the platform Amazon Mechanical Turk

psxng1nottinghamacuk

Neha_Gi_ji

Research amp Research teams

Research Studies

bull Qualitative study of Turkers on online forums like mturkforum and Turker Nation by Dr David Martin XRCE

bull Ethnographic Studies of Turkers in their respective homes and places of work in India by Neha Gupta UoN

Research teams

Xerox

bull David Martin Jacki OrsquoNeill (Now MSR India) Ben Hanrahan (Now PSU)

psxng1nottinghamacuk

Neha_Gi_ji

Nottingham

bull Andy Crabtree

Tom Rodden

Matthias Hirth

Matthias Hirth 48 48

About Myself

Matthias Hirth

Research Assistant at University of Wuumlrzburg

httpmatthias-hirthcom

matthiashirthinformatikuni-wuerzburgde

Crowdsourcing

Platform analysis

Platform optimization

Human factors

Use cases

Online Social Networks

Collaboration networks

Structural differences of OSNs

Application of OSN relationships

Interactive applicationsGaming

Challenges of real time interactions

Serious gamingGamification

E-Sports broadcastinglive streaming

Quality of Experience

Monitoring techniques

Quantification techniques

(Crowd-bases) test design

Social interactions

Test methodology Incentive design

Use case

Measurement technique

Identification of

qualification

Matthias Hirth 49 49

Improving Crowdsourcing Technology and the Way We Use it

Crowdsourcing technology

How to transfer well known lab procedures into an uncontrolled

crowdsourcing environment

Which crowdsourcing platform features are required to conduct

successful experiments but are not present yet

Which technical developments (can) foster the usage of

crowdsourcing instead of lab experiments

How can technological developments be used for novel incentive

mechanisms

Usage of Crowdsourcing

How can we conduct reproducible experiments in diverse

crowdsourcing environment

How can we raise the awareness of crowdsourcing employers for the

individuals behind the crowd

Tobias Hoszligfeld

The crowd community in the lab

Tobias Hoszligfeld wwwmaswiwiuni-duede

bull Background Tobias Hoszligfeld

ndash 2004-2014 Research group ldquoFuture Internet Applicationsrdquo Chair of Communication Networks Univ of Wuumlrzburg

ndash Since Sep 2014 Chair of Modeling of Adaptive Systems (head) University of Duisburg-Essen

bull Interests Application of crowdsourcing

ndash User studies on Quality of Experience (QoE) for Internet apps eg video QoE in the crowd amp in the lab methodology reliability statisticshellip [2]

ndash Mobile crowdsourcing network and context measurements eg Wifi offloading incentives privacy recommendation mechanisms etc [1]

bull Interests The crowd community

ndash Crowd = Human beings Social network wisdom of the crowd community

ndash UnderstandingUtilizing the crowd community for successful human computation For which jobs can social interaction improve the quality of work Classification

ndash How can we deploy such a system where interaction is triggered

ndash System design Task design interaction mechanisms incentives

22 Nov 2015

[1] httpdropsdagstuhldeopusvolltexte20134354

[2] httpswww3informatikuni-wuerzburgdeqoewikiqualinetcrowd

Challenges Example of Crowdsourced QoE

Tobias Hoszligfeld wwwmaswiwiuni-duede

task design

incentives

Internet

crowd

community

anonymous

remote

subjects

Internet delivery of

test data and user

response

QoE study bull testing methodology bull QoE influence factors

under investigation bull additional QoE factors bull proper subset of items bull hellip

crowdsourcing setting bull short task duration bull instructions amp training bull feedback amp questions bull hidden influence factors bull reliability of users bull hellip

heterogeneous devices

and Internet access

implementation extrinsic vs intrinsic

motivation

Processing of data

Technical aspects bull server implementation bull client application amp

compatibility bull desired test conditions bull context monitoring bull hellip

test items

lab

scientific methodologies

22 Nov 2015

Jason Jacques

Jason Jacques supervised by Per Ola Kristensson

T

Crowdsourcing Code Perception

jtj21camacuk peopledscamacukjtj21

jtjacquesgmailcom jsonjcouk

PhD Working Title

science

Crowdsourcing Interests W

hat

W

ho

Ho

w

Wh

y jtj21camacuk jsonjcouk science

Radu Jianu

Radu Jianu Assistant Professor

Florida International University (PhD Brown University lsquo12)

Interests

Applications of eye-tracking in visualization tracking what people look at rather than where they look

bull Data intensive large scale eye-tracking experiments bull Gaze-contingent visualizations bull Gaze-driven data recommendation and big data curation

Automating user studies in data visualizations (on next page)

Automating user studies

1 Software designs that can (semi)automate a broad range of quantitative evaluations by leveraging

ndash Crowdsourcing ndash Benchmark data and task taxonomies ndash Automated result interpretation

2 Moving user evaluation from the end to within the software design and implementation process as a means to choose between competitive visual designs

Christian Keimel

copy IRT

Where

What

bull Hybrid broadcast broadband technologies

bull Big Data in broadcasting environment and content-creation

bull Audio-visual QoE assessment (with crowdsourcing)

About me

Christian Keimel Slide 60

Lecturer for Digital Broadcast Engineering

Research Engineer for Media Services and Platforms

copy IRT

Crowdsourcing in large scale optimisation

The human component of Big Data

Moving crowdsourcing to lean-back devices

Crowdsourcing from the couch

Crowdsourced experiments emulating the laboratory

Is it time for new methodologies ignoring the lab

Questionshellip

Christian Keimel Slide 61

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 30: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Darren Edwards Swansea University

Cognitive psychology - categorization

Health and clinical psychology ndash applications of categorization on clinical populations (eg autism)

Crowdsourcing and psychological interventions

Recent validity studies confirm the use of simple psychological tasks such as decision making

Recent attempts have now explored more complex tasks such as psychological interventions (eg mindfulness)

Present studies now explore more complex forms of mindfulness

What are the limitations of using crowdsourced populations online such as when using psychological interventions or categorization

Sebastian Egger

Engineering

Technical high school

bull Telematics

electrotechnical enginieering telecommunications

bull Signal Processing for UWB systems

Sociology

Theory of human societies

Interaction amp collaboration behaviour of groups

Measurement of human behaviour

Currently Human Computer Interaction

Social amp collaborative aspects of technology use

33 26112015

Sebastian Egger

Motivational aspects of crowdworkers

CS as a collaborative work setting

Optimal Setup of CS experiments

Questions (max Nr of questions max duration)

Gamified reliability checks

Scale designs

CS vs Lab whatlsquos the ground truth

New application domains of CS

34 26112015

Interest in Crowdsourcing

Sara Fabrikant

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

humanmdashsystem (visuo-spatial) interface

spatio-temporal analytics

spatialization human navigation

multivariate spatial analysis etc)

interface design

large amp small interactive (map) displays

(ie desktop mobile VR etc)

empirical evaluations

(ie eye tracking EDA EEG etc) [httpwwwunitsevenconz]

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

human sensing observatory set up

controlled laboratory | in-situ (messy) real world setting

semi|automatic data collection of various human behavioral data streams

ie psycho-physiological perceptual cognitive and sensory-motor behaviors (through crowdsourcing )

Brian Fisher

My double research life

Brian Fisher

bull PhD in Experimental Psych

bull Postdoc w Cognitive Science

society founder amp president

bull Psychonomics Fellow

bull VIS-related symposia at Cogsci

amp APS papers on cogsci of

interaction

bull Fuzzy-logicBayes models

bull Postdoc funded by Inst for Robotics

and AI

bull VAST SC VEC VACCINE

bull Co-organized Dagstuhl ldquoInteraction

with Information for Visual Reasoningrdquo

Cogsci-based papers at VIS CHI

BELIV

My Plan

VIS

Me Psychological Science

Ujwal Gadiraju

p = 0055111

Crowd Worker

Behavior

Microtask

Design

Crowdsourcing Paradigm Quality amp Effectiveness

UJWAL GADIRAJU

Incentivization Gamification

Neha Gupta

Getting to know the crowd

Presented at Dagstuhl by

Dr David Martin amp Neha Gupta

Neha Gupta

PhD Student School of Computer Science University of Nottingham UK

Investigating crowdwork through the platform Amazon Mechanical Turk

psxng1nottinghamacuk

Neha_Gi_ji

Research amp Research teams

Research Studies

bull Qualitative study of Turkers on online forums like mturkforum and Turker Nation by Dr David Martin XRCE

bull Ethnographic Studies of Turkers in their respective homes and places of work in India by Neha Gupta UoN

Research teams

Xerox

bull David Martin Jacki OrsquoNeill (Now MSR India) Ben Hanrahan (Now PSU)

psxng1nottinghamacuk

Neha_Gi_ji

Nottingham

bull Andy Crabtree

Tom Rodden

Matthias Hirth

Matthias Hirth 48 48

About Myself

Matthias Hirth

Research Assistant at University of Wuumlrzburg

httpmatthias-hirthcom

matthiashirthinformatikuni-wuerzburgde

Crowdsourcing

Platform analysis

Platform optimization

Human factors

Use cases

Online Social Networks

Collaboration networks

Structural differences of OSNs

Application of OSN relationships

Interactive applicationsGaming

Challenges of real time interactions

Serious gamingGamification

E-Sports broadcastinglive streaming

Quality of Experience

Monitoring techniques

Quantification techniques

(Crowd-bases) test design

Social interactions

Test methodology Incentive design

Use case

Measurement technique

Identification of

qualification

Matthias Hirth 49 49

Improving Crowdsourcing Technology and the Way We Use it

Crowdsourcing technology

How to transfer well known lab procedures into an uncontrolled

crowdsourcing environment

Which crowdsourcing platform features are required to conduct

successful experiments but are not present yet

Which technical developments (can) foster the usage of

crowdsourcing instead of lab experiments

How can technological developments be used for novel incentive

mechanisms

Usage of Crowdsourcing

How can we conduct reproducible experiments in diverse

crowdsourcing environment

How can we raise the awareness of crowdsourcing employers for the

individuals behind the crowd

Tobias Hoszligfeld

The crowd community in the lab

Tobias Hoszligfeld wwwmaswiwiuni-duede

bull Background Tobias Hoszligfeld

ndash 2004-2014 Research group ldquoFuture Internet Applicationsrdquo Chair of Communication Networks Univ of Wuumlrzburg

ndash Since Sep 2014 Chair of Modeling of Adaptive Systems (head) University of Duisburg-Essen

bull Interests Application of crowdsourcing

ndash User studies on Quality of Experience (QoE) for Internet apps eg video QoE in the crowd amp in the lab methodology reliability statisticshellip [2]

ndash Mobile crowdsourcing network and context measurements eg Wifi offloading incentives privacy recommendation mechanisms etc [1]

bull Interests The crowd community

ndash Crowd = Human beings Social network wisdom of the crowd community

ndash UnderstandingUtilizing the crowd community for successful human computation For which jobs can social interaction improve the quality of work Classification

ndash How can we deploy such a system where interaction is triggered

ndash System design Task design interaction mechanisms incentives

22 Nov 2015

[1] httpdropsdagstuhldeopusvolltexte20134354

[2] httpswww3informatikuni-wuerzburgdeqoewikiqualinetcrowd

Challenges Example of Crowdsourced QoE

Tobias Hoszligfeld wwwmaswiwiuni-duede

task design

incentives

Internet

crowd

community

anonymous

remote

subjects

Internet delivery of

test data and user

response

QoE study bull testing methodology bull QoE influence factors

under investigation bull additional QoE factors bull proper subset of items bull hellip

crowdsourcing setting bull short task duration bull instructions amp training bull feedback amp questions bull hidden influence factors bull reliability of users bull hellip

heterogeneous devices

and Internet access

implementation extrinsic vs intrinsic

motivation

Processing of data

Technical aspects bull server implementation bull client application amp

compatibility bull desired test conditions bull context monitoring bull hellip

test items

lab

scientific methodologies

22 Nov 2015

Jason Jacques

Jason Jacques supervised by Per Ola Kristensson

T

Crowdsourcing Code Perception

jtj21camacuk peopledscamacukjtj21

jtjacquesgmailcom jsonjcouk

PhD Working Title

science

Crowdsourcing Interests W

hat

W

ho

Ho

w

Wh

y jtj21camacuk jsonjcouk science

Radu Jianu

Radu Jianu Assistant Professor

Florida International University (PhD Brown University lsquo12)

Interests

Applications of eye-tracking in visualization tracking what people look at rather than where they look

bull Data intensive large scale eye-tracking experiments bull Gaze-contingent visualizations bull Gaze-driven data recommendation and big data curation

Automating user studies in data visualizations (on next page)

Automating user studies

1 Software designs that can (semi)automate a broad range of quantitative evaluations by leveraging

ndash Crowdsourcing ndash Benchmark data and task taxonomies ndash Automated result interpretation

2 Moving user evaluation from the end to within the software design and implementation process as a means to choose between competitive visual designs

Christian Keimel

copy IRT

Where

What

bull Hybrid broadcast broadband technologies

bull Big Data in broadcasting environment and content-creation

bull Audio-visual QoE assessment (with crowdsourcing)

About me

Christian Keimel Slide 60

Lecturer for Digital Broadcast Engineering

Research Engineer for Media Services and Platforms

copy IRT

Crowdsourcing in large scale optimisation

The human component of Big Data

Moving crowdsourcing to lean-back devices

Crowdsourcing from the couch

Crowdsourced experiments emulating the laboratory

Is it time for new methodologies ignoring the lab

Questionshellip

Christian Keimel Slide 61

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 31: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Crowdsourcing and psychological interventions

Recent validity studies confirm the use of simple psychological tasks such as decision making

Recent attempts have now explored more complex tasks such as psychological interventions (eg mindfulness)

Present studies now explore more complex forms of mindfulness

What are the limitations of using crowdsourced populations online such as when using psychological interventions or categorization

Sebastian Egger

Engineering

Technical high school

bull Telematics

electrotechnical enginieering telecommunications

bull Signal Processing for UWB systems

Sociology

Theory of human societies

Interaction amp collaboration behaviour of groups

Measurement of human behaviour

Currently Human Computer Interaction

Social amp collaborative aspects of technology use

33 26112015

Sebastian Egger

Motivational aspects of crowdworkers

CS as a collaborative work setting

Optimal Setup of CS experiments

Questions (max Nr of questions max duration)

Gamified reliability checks

Scale designs

CS vs Lab whatlsquos the ground truth

New application domains of CS

34 26112015

Interest in Crowdsourcing

Sara Fabrikant

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

humanmdashsystem (visuo-spatial) interface

spatio-temporal analytics

spatialization human navigation

multivariate spatial analysis etc)

interface design

large amp small interactive (map) displays

(ie desktop mobile VR etc)

empirical evaluations

(ie eye tracking EDA EEG etc) [httpwwwunitsevenconz]

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

human sensing observatory set up

controlled laboratory | in-situ (messy) real world setting

semi|automatic data collection of various human behavioral data streams

ie psycho-physiological perceptual cognitive and sensory-motor behaviors (through crowdsourcing )

Brian Fisher

My double research life

Brian Fisher

bull PhD in Experimental Psych

bull Postdoc w Cognitive Science

society founder amp president

bull Psychonomics Fellow

bull VIS-related symposia at Cogsci

amp APS papers on cogsci of

interaction

bull Fuzzy-logicBayes models

bull Postdoc funded by Inst for Robotics

and AI

bull VAST SC VEC VACCINE

bull Co-organized Dagstuhl ldquoInteraction

with Information for Visual Reasoningrdquo

Cogsci-based papers at VIS CHI

BELIV

My Plan

VIS

Me Psychological Science

Ujwal Gadiraju

p = 0055111

Crowd Worker

Behavior

Microtask

Design

Crowdsourcing Paradigm Quality amp Effectiveness

UJWAL GADIRAJU

Incentivization Gamification

Neha Gupta

Getting to know the crowd

Presented at Dagstuhl by

Dr David Martin amp Neha Gupta

Neha Gupta

PhD Student School of Computer Science University of Nottingham UK

Investigating crowdwork through the platform Amazon Mechanical Turk

psxng1nottinghamacuk

Neha_Gi_ji

Research amp Research teams

Research Studies

bull Qualitative study of Turkers on online forums like mturkforum and Turker Nation by Dr David Martin XRCE

bull Ethnographic Studies of Turkers in their respective homes and places of work in India by Neha Gupta UoN

Research teams

Xerox

bull David Martin Jacki OrsquoNeill (Now MSR India) Ben Hanrahan (Now PSU)

psxng1nottinghamacuk

Neha_Gi_ji

Nottingham

bull Andy Crabtree

Tom Rodden

Matthias Hirth

Matthias Hirth 48 48

About Myself

Matthias Hirth

Research Assistant at University of Wuumlrzburg

httpmatthias-hirthcom

matthiashirthinformatikuni-wuerzburgde

Crowdsourcing

Platform analysis

Platform optimization

Human factors

Use cases

Online Social Networks

Collaboration networks

Structural differences of OSNs

Application of OSN relationships

Interactive applicationsGaming

Challenges of real time interactions

Serious gamingGamification

E-Sports broadcastinglive streaming

Quality of Experience

Monitoring techniques

Quantification techniques

(Crowd-bases) test design

Social interactions

Test methodology Incentive design

Use case

Measurement technique

Identification of

qualification

Matthias Hirth 49 49

Improving Crowdsourcing Technology and the Way We Use it

Crowdsourcing technology

How to transfer well known lab procedures into an uncontrolled

crowdsourcing environment

Which crowdsourcing platform features are required to conduct

successful experiments but are not present yet

Which technical developments (can) foster the usage of

crowdsourcing instead of lab experiments

How can technological developments be used for novel incentive

mechanisms

Usage of Crowdsourcing

How can we conduct reproducible experiments in diverse

crowdsourcing environment

How can we raise the awareness of crowdsourcing employers for the

individuals behind the crowd

Tobias Hoszligfeld

The crowd community in the lab

Tobias Hoszligfeld wwwmaswiwiuni-duede

bull Background Tobias Hoszligfeld

ndash 2004-2014 Research group ldquoFuture Internet Applicationsrdquo Chair of Communication Networks Univ of Wuumlrzburg

ndash Since Sep 2014 Chair of Modeling of Adaptive Systems (head) University of Duisburg-Essen

bull Interests Application of crowdsourcing

ndash User studies on Quality of Experience (QoE) for Internet apps eg video QoE in the crowd amp in the lab methodology reliability statisticshellip [2]

ndash Mobile crowdsourcing network and context measurements eg Wifi offloading incentives privacy recommendation mechanisms etc [1]

bull Interests The crowd community

ndash Crowd = Human beings Social network wisdom of the crowd community

ndash UnderstandingUtilizing the crowd community for successful human computation For which jobs can social interaction improve the quality of work Classification

ndash How can we deploy such a system where interaction is triggered

ndash System design Task design interaction mechanisms incentives

22 Nov 2015

[1] httpdropsdagstuhldeopusvolltexte20134354

[2] httpswww3informatikuni-wuerzburgdeqoewikiqualinetcrowd

Challenges Example of Crowdsourced QoE

Tobias Hoszligfeld wwwmaswiwiuni-duede

task design

incentives

Internet

crowd

community

anonymous

remote

subjects

Internet delivery of

test data and user

response

QoE study bull testing methodology bull QoE influence factors

under investigation bull additional QoE factors bull proper subset of items bull hellip

crowdsourcing setting bull short task duration bull instructions amp training bull feedback amp questions bull hidden influence factors bull reliability of users bull hellip

heterogeneous devices

and Internet access

implementation extrinsic vs intrinsic

motivation

Processing of data

Technical aspects bull server implementation bull client application amp

compatibility bull desired test conditions bull context monitoring bull hellip

test items

lab

scientific methodologies

22 Nov 2015

Jason Jacques

Jason Jacques supervised by Per Ola Kristensson

T

Crowdsourcing Code Perception

jtj21camacuk peopledscamacukjtj21

jtjacquesgmailcom jsonjcouk

PhD Working Title

science

Crowdsourcing Interests W

hat

W

ho

Ho

w

Wh

y jtj21camacuk jsonjcouk science

Radu Jianu

Radu Jianu Assistant Professor

Florida International University (PhD Brown University lsquo12)

Interests

Applications of eye-tracking in visualization tracking what people look at rather than where they look

bull Data intensive large scale eye-tracking experiments bull Gaze-contingent visualizations bull Gaze-driven data recommendation and big data curation

Automating user studies in data visualizations (on next page)

Automating user studies

1 Software designs that can (semi)automate a broad range of quantitative evaluations by leveraging

ndash Crowdsourcing ndash Benchmark data and task taxonomies ndash Automated result interpretation

2 Moving user evaluation from the end to within the software design and implementation process as a means to choose between competitive visual designs

Christian Keimel

copy IRT

Where

What

bull Hybrid broadcast broadband technologies

bull Big Data in broadcasting environment and content-creation

bull Audio-visual QoE assessment (with crowdsourcing)

About me

Christian Keimel Slide 60

Lecturer for Digital Broadcast Engineering

Research Engineer for Media Services and Platforms

copy IRT

Crowdsourcing in large scale optimisation

The human component of Big Data

Moving crowdsourcing to lean-back devices

Crowdsourcing from the couch

Crowdsourced experiments emulating the laboratory

Is it time for new methodologies ignoring the lab

Questionshellip

Christian Keimel Slide 61

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 32: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Sebastian Egger

Engineering

Technical high school

bull Telematics

electrotechnical enginieering telecommunications

bull Signal Processing for UWB systems

Sociology

Theory of human societies

Interaction amp collaboration behaviour of groups

Measurement of human behaviour

Currently Human Computer Interaction

Social amp collaborative aspects of technology use

33 26112015

Sebastian Egger

Motivational aspects of crowdworkers

CS as a collaborative work setting

Optimal Setup of CS experiments

Questions (max Nr of questions max duration)

Gamified reliability checks

Scale designs

CS vs Lab whatlsquos the ground truth

New application domains of CS

34 26112015

Interest in Crowdsourcing

Sara Fabrikant

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

humanmdashsystem (visuo-spatial) interface

spatio-temporal analytics

spatialization human navigation

multivariate spatial analysis etc)

interface design

large amp small interactive (map) displays

(ie desktop mobile VR etc)

empirical evaluations

(ie eye tracking EDA EEG etc) [httpwwwunitsevenconz]

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

human sensing observatory set up

controlled laboratory | in-situ (messy) real world setting

semi|automatic data collection of various human behavioral data streams

ie psycho-physiological perceptual cognitive and sensory-motor behaviors (through crowdsourcing )

Brian Fisher

My double research life

Brian Fisher

bull PhD in Experimental Psych

bull Postdoc w Cognitive Science

society founder amp president

bull Psychonomics Fellow

bull VIS-related symposia at Cogsci

amp APS papers on cogsci of

interaction

bull Fuzzy-logicBayes models

bull Postdoc funded by Inst for Robotics

and AI

bull VAST SC VEC VACCINE

bull Co-organized Dagstuhl ldquoInteraction

with Information for Visual Reasoningrdquo

Cogsci-based papers at VIS CHI

BELIV

My Plan

VIS

Me Psychological Science

Ujwal Gadiraju

p = 0055111

Crowd Worker

Behavior

Microtask

Design

Crowdsourcing Paradigm Quality amp Effectiveness

UJWAL GADIRAJU

Incentivization Gamification

Neha Gupta

Getting to know the crowd

Presented at Dagstuhl by

Dr David Martin amp Neha Gupta

Neha Gupta

PhD Student School of Computer Science University of Nottingham UK

Investigating crowdwork through the platform Amazon Mechanical Turk

psxng1nottinghamacuk

Neha_Gi_ji

Research amp Research teams

Research Studies

bull Qualitative study of Turkers on online forums like mturkforum and Turker Nation by Dr David Martin XRCE

bull Ethnographic Studies of Turkers in their respective homes and places of work in India by Neha Gupta UoN

Research teams

Xerox

bull David Martin Jacki OrsquoNeill (Now MSR India) Ben Hanrahan (Now PSU)

psxng1nottinghamacuk

Neha_Gi_ji

Nottingham

bull Andy Crabtree

Tom Rodden

Matthias Hirth

Matthias Hirth 48 48

About Myself

Matthias Hirth

Research Assistant at University of Wuumlrzburg

httpmatthias-hirthcom

matthiashirthinformatikuni-wuerzburgde

Crowdsourcing

Platform analysis

Platform optimization

Human factors

Use cases

Online Social Networks

Collaboration networks

Structural differences of OSNs

Application of OSN relationships

Interactive applicationsGaming

Challenges of real time interactions

Serious gamingGamification

E-Sports broadcastinglive streaming

Quality of Experience

Monitoring techniques

Quantification techniques

(Crowd-bases) test design

Social interactions

Test methodology Incentive design

Use case

Measurement technique

Identification of

qualification

Matthias Hirth 49 49

Improving Crowdsourcing Technology and the Way We Use it

Crowdsourcing technology

How to transfer well known lab procedures into an uncontrolled

crowdsourcing environment

Which crowdsourcing platform features are required to conduct

successful experiments but are not present yet

Which technical developments (can) foster the usage of

crowdsourcing instead of lab experiments

How can technological developments be used for novel incentive

mechanisms

Usage of Crowdsourcing

How can we conduct reproducible experiments in diverse

crowdsourcing environment

How can we raise the awareness of crowdsourcing employers for the

individuals behind the crowd

Tobias Hoszligfeld

The crowd community in the lab

Tobias Hoszligfeld wwwmaswiwiuni-duede

bull Background Tobias Hoszligfeld

ndash 2004-2014 Research group ldquoFuture Internet Applicationsrdquo Chair of Communication Networks Univ of Wuumlrzburg

ndash Since Sep 2014 Chair of Modeling of Adaptive Systems (head) University of Duisburg-Essen

bull Interests Application of crowdsourcing

ndash User studies on Quality of Experience (QoE) for Internet apps eg video QoE in the crowd amp in the lab methodology reliability statisticshellip [2]

ndash Mobile crowdsourcing network and context measurements eg Wifi offloading incentives privacy recommendation mechanisms etc [1]

bull Interests The crowd community

ndash Crowd = Human beings Social network wisdom of the crowd community

ndash UnderstandingUtilizing the crowd community for successful human computation For which jobs can social interaction improve the quality of work Classification

ndash How can we deploy such a system where interaction is triggered

ndash System design Task design interaction mechanisms incentives

22 Nov 2015

[1] httpdropsdagstuhldeopusvolltexte20134354

[2] httpswww3informatikuni-wuerzburgdeqoewikiqualinetcrowd

Challenges Example of Crowdsourced QoE

Tobias Hoszligfeld wwwmaswiwiuni-duede

task design

incentives

Internet

crowd

community

anonymous

remote

subjects

Internet delivery of

test data and user

response

QoE study bull testing methodology bull QoE influence factors

under investigation bull additional QoE factors bull proper subset of items bull hellip

crowdsourcing setting bull short task duration bull instructions amp training bull feedback amp questions bull hidden influence factors bull reliability of users bull hellip

heterogeneous devices

and Internet access

implementation extrinsic vs intrinsic

motivation

Processing of data

Technical aspects bull server implementation bull client application amp

compatibility bull desired test conditions bull context monitoring bull hellip

test items

lab

scientific methodologies

22 Nov 2015

Jason Jacques

Jason Jacques supervised by Per Ola Kristensson

T

Crowdsourcing Code Perception

jtj21camacuk peopledscamacukjtj21

jtjacquesgmailcom jsonjcouk

PhD Working Title

science

Crowdsourcing Interests W

hat

W

ho

Ho

w

Wh

y jtj21camacuk jsonjcouk science

Radu Jianu

Radu Jianu Assistant Professor

Florida International University (PhD Brown University lsquo12)

Interests

Applications of eye-tracking in visualization tracking what people look at rather than where they look

bull Data intensive large scale eye-tracking experiments bull Gaze-contingent visualizations bull Gaze-driven data recommendation and big data curation

Automating user studies in data visualizations (on next page)

Automating user studies

1 Software designs that can (semi)automate a broad range of quantitative evaluations by leveraging

ndash Crowdsourcing ndash Benchmark data and task taxonomies ndash Automated result interpretation

2 Moving user evaluation from the end to within the software design and implementation process as a means to choose between competitive visual designs

Christian Keimel

copy IRT

Where

What

bull Hybrid broadcast broadband technologies

bull Big Data in broadcasting environment and content-creation

bull Audio-visual QoE assessment (with crowdsourcing)

About me

Christian Keimel Slide 60

Lecturer for Digital Broadcast Engineering

Research Engineer for Media Services and Platforms

copy IRT

Crowdsourcing in large scale optimisation

The human component of Big Data

Moving crowdsourcing to lean-back devices

Crowdsourcing from the couch

Crowdsourced experiments emulating the laboratory

Is it time for new methodologies ignoring the lab

Questionshellip

Christian Keimel Slide 61

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 33: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Engineering

Technical high school

bull Telematics

electrotechnical enginieering telecommunications

bull Signal Processing for UWB systems

Sociology

Theory of human societies

Interaction amp collaboration behaviour of groups

Measurement of human behaviour

Currently Human Computer Interaction

Social amp collaborative aspects of technology use

33 26112015

Sebastian Egger

Motivational aspects of crowdworkers

CS as a collaborative work setting

Optimal Setup of CS experiments

Questions (max Nr of questions max duration)

Gamified reliability checks

Scale designs

CS vs Lab whatlsquos the ground truth

New application domains of CS

34 26112015

Interest in Crowdsourcing

Sara Fabrikant

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

humanmdashsystem (visuo-spatial) interface

spatio-temporal analytics

spatialization human navigation

multivariate spatial analysis etc)

interface design

large amp small interactive (map) displays

(ie desktop mobile VR etc)

empirical evaluations

(ie eye tracking EDA EEG etc) [httpwwwunitsevenconz]

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

human sensing observatory set up

controlled laboratory | in-situ (messy) real world setting

semi|automatic data collection of various human behavioral data streams

ie psycho-physiological perceptual cognitive and sensory-motor behaviors (through crowdsourcing )

Brian Fisher

My double research life

Brian Fisher

bull PhD in Experimental Psych

bull Postdoc w Cognitive Science

society founder amp president

bull Psychonomics Fellow

bull VIS-related symposia at Cogsci

amp APS papers on cogsci of

interaction

bull Fuzzy-logicBayes models

bull Postdoc funded by Inst for Robotics

and AI

bull VAST SC VEC VACCINE

bull Co-organized Dagstuhl ldquoInteraction

with Information for Visual Reasoningrdquo

Cogsci-based papers at VIS CHI

BELIV

My Plan

VIS

Me Psychological Science

Ujwal Gadiraju

p = 0055111

Crowd Worker

Behavior

Microtask

Design

Crowdsourcing Paradigm Quality amp Effectiveness

UJWAL GADIRAJU

Incentivization Gamification

Neha Gupta

Getting to know the crowd

Presented at Dagstuhl by

Dr David Martin amp Neha Gupta

Neha Gupta

PhD Student School of Computer Science University of Nottingham UK

Investigating crowdwork through the platform Amazon Mechanical Turk

psxng1nottinghamacuk

Neha_Gi_ji

Research amp Research teams

Research Studies

bull Qualitative study of Turkers on online forums like mturkforum and Turker Nation by Dr David Martin XRCE

bull Ethnographic Studies of Turkers in their respective homes and places of work in India by Neha Gupta UoN

Research teams

Xerox

bull David Martin Jacki OrsquoNeill (Now MSR India) Ben Hanrahan (Now PSU)

psxng1nottinghamacuk

Neha_Gi_ji

Nottingham

bull Andy Crabtree

Tom Rodden

Matthias Hirth

Matthias Hirth 48 48

About Myself

Matthias Hirth

Research Assistant at University of Wuumlrzburg

httpmatthias-hirthcom

matthiashirthinformatikuni-wuerzburgde

Crowdsourcing

Platform analysis

Platform optimization

Human factors

Use cases

Online Social Networks

Collaboration networks

Structural differences of OSNs

Application of OSN relationships

Interactive applicationsGaming

Challenges of real time interactions

Serious gamingGamification

E-Sports broadcastinglive streaming

Quality of Experience

Monitoring techniques

Quantification techniques

(Crowd-bases) test design

Social interactions

Test methodology Incentive design

Use case

Measurement technique

Identification of

qualification

Matthias Hirth 49 49

Improving Crowdsourcing Technology and the Way We Use it

Crowdsourcing technology

How to transfer well known lab procedures into an uncontrolled

crowdsourcing environment

Which crowdsourcing platform features are required to conduct

successful experiments but are not present yet

Which technical developments (can) foster the usage of

crowdsourcing instead of lab experiments

How can technological developments be used for novel incentive

mechanisms

Usage of Crowdsourcing

How can we conduct reproducible experiments in diverse

crowdsourcing environment

How can we raise the awareness of crowdsourcing employers for the

individuals behind the crowd

Tobias Hoszligfeld

The crowd community in the lab

Tobias Hoszligfeld wwwmaswiwiuni-duede

bull Background Tobias Hoszligfeld

ndash 2004-2014 Research group ldquoFuture Internet Applicationsrdquo Chair of Communication Networks Univ of Wuumlrzburg

ndash Since Sep 2014 Chair of Modeling of Adaptive Systems (head) University of Duisburg-Essen

bull Interests Application of crowdsourcing

ndash User studies on Quality of Experience (QoE) for Internet apps eg video QoE in the crowd amp in the lab methodology reliability statisticshellip [2]

ndash Mobile crowdsourcing network and context measurements eg Wifi offloading incentives privacy recommendation mechanisms etc [1]

bull Interests The crowd community

ndash Crowd = Human beings Social network wisdom of the crowd community

ndash UnderstandingUtilizing the crowd community for successful human computation For which jobs can social interaction improve the quality of work Classification

ndash How can we deploy such a system where interaction is triggered

ndash System design Task design interaction mechanisms incentives

22 Nov 2015

[1] httpdropsdagstuhldeopusvolltexte20134354

[2] httpswww3informatikuni-wuerzburgdeqoewikiqualinetcrowd

Challenges Example of Crowdsourced QoE

Tobias Hoszligfeld wwwmaswiwiuni-duede

task design

incentives

Internet

crowd

community

anonymous

remote

subjects

Internet delivery of

test data and user

response

QoE study bull testing methodology bull QoE influence factors

under investigation bull additional QoE factors bull proper subset of items bull hellip

crowdsourcing setting bull short task duration bull instructions amp training bull feedback amp questions bull hidden influence factors bull reliability of users bull hellip

heterogeneous devices

and Internet access

implementation extrinsic vs intrinsic

motivation

Processing of data

Technical aspects bull server implementation bull client application amp

compatibility bull desired test conditions bull context monitoring bull hellip

test items

lab

scientific methodologies

22 Nov 2015

Jason Jacques

Jason Jacques supervised by Per Ola Kristensson

T

Crowdsourcing Code Perception

jtj21camacuk peopledscamacukjtj21

jtjacquesgmailcom jsonjcouk

PhD Working Title

science

Crowdsourcing Interests W

hat

W

ho

Ho

w

Wh

y jtj21camacuk jsonjcouk science

Radu Jianu

Radu Jianu Assistant Professor

Florida International University (PhD Brown University lsquo12)

Interests

Applications of eye-tracking in visualization tracking what people look at rather than where they look

bull Data intensive large scale eye-tracking experiments bull Gaze-contingent visualizations bull Gaze-driven data recommendation and big data curation

Automating user studies in data visualizations (on next page)

Automating user studies

1 Software designs that can (semi)automate a broad range of quantitative evaluations by leveraging

ndash Crowdsourcing ndash Benchmark data and task taxonomies ndash Automated result interpretation

2 Moving user evaluation from the end to within the software design and implementation process as a means to choose between competitive visual designs

Christian Keimel

copy IRT

Where

What

bull Hybrid broadcast broadband technologies

bull Big Data in broadcasting environment and content-creation

bull Audio-visual QoE assessment (with crowdsourcing)

About me

Christian Keimel Slide 60

Lecturer for Digital Broadcast Engineering

Research Engineer for Media Services and Platforms

copy IRT

Crowdsourcing in large scale optimisation

The human component of Big Data

Moving crowdsourcing to lean-back devices

Crowdsourcing from the couch

Crowdsourced experiments emulating the laboratory

Is it time for new methodologies ignoring the lab

Questionshellip

Christian Keimel Slide 61

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 34: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Motivational aspects of crowdworkers

CS as a collaborative work setting

Optimal Setup of CS experiments

Questions (max Nr of questions max duration)

Gamified reliability checks

Scale designs

CS vs Lab whatlsquos the ground truth

New application domains of CS

34 26112015

Interest in Crowdsourcing

Sara Fabrikant

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

humanmdashsystem (visuo-spatial) interface

spatio-temporal analytics

spatialization human navigation

multivariate spatial analysis etc)

interface design

large amp small interactive (map) displays

(ie desktop mobile VR etc)

empirical evaluations

(ie eye tracking EDA EEG etc) [httpwwwunitsevenconz]

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

human sensing observatory set up

controlled laboratory | in-situ (messy) real world setting

semi|automatic data collection of various human behavioral data streams

ie psycho-physiological perceptual cognitive and sensory-motor behaviors (through crowdsourcing )

Brian Fisher

My double research life

Brian Fisher

bull PhD in Experimental Psych

bull Postdoc w Cognitive Science

society founder amp president

bull Psychonomics Fellow

bull VIS-related symposia at Cogsci

amp APS papers on cogsci of

interaction

bull Fuzzy-logicBayes models

bull Postdoc funded by Inst for Robotics

and AI

bull VAST SC VEC VACCINE

bull Co-organized Dagstuhl ldquoInteraction

with Information for Visual Reasoningrdquo

Cogsci-based papers at VIS CHI

BELIV

My Plan

VIS

Me Psychological Science

Ujwal Gadiraju

p = 0055111

Crowd Worker

Behavior

Microtask

Design

Crowdsourcing Paradigm Quality amp Effectiveness

UJWAL GADIRAJU

Incentivization Gamification

Neha Gupta

Getting to know the crowd

Presented at Dagstuhl by

Dr David Martin amp Neha Gupta

Neha Gupta

PhD Student School of Computer Science University of Nottingham UK

Investigating crowdwork through the platform Amazon Mechanical Turk

psxng1nottinghamacuk

Neha_Gi_ji

Research amp Research teams

Research Studies

bull Qualitative study of Turkers on online forums like mturkforum and Turker Nation by Dr David Martin XRCE

bull Ethnographic Studies of Turkers in their respective homes and places of work in India by Neha Gupta UoN

Research teams

Xerox

bull David Martin Jacki OrsquoNeill (Now MSR India) Ben Hanrahan (Now PSU)

psxng1nottinghamacuk

Neha_Gi_ji

Nottingham

bull Andy Crabtree

Tom Rodden

Matthias Hirth

Matthias Hirth 48 48

About Myself

Matthias Hirth

Research Assistant at University of Wuumlrzburg

httpmatthias-hirthcom

matthiashirthinformatikuni-wuerzburgde

Crowdsourcing

Platform analysis

Platform optimization

Human factors

Use cases

Online Social Networks

Collaboration networks

Structural differences of OSNs

Application of OSN relationships

Interactive applicationsGaming

Challenges of real time interactions

Serious gamingGamification

E-Sports broadcastinglive streaming

Quality of Experience

Monitoring techniques

Quantification techniques

(Crowd-bases) test design

Social interactions

Test methodology Incentive design

Use case

Measurement technique

Identification of

qualification

Matthias Hirth 49 49

Improving Crowdsourcing Technology and the Way We Use it

Crowdsourcing technology

How to transfer well known lab procedures into an uncontrolled

crowdsourcing environment

Which crowdsourcing platform features are required to conduct

successful experiments but are not present yet

Which technical developments (can) foster the usage of

crowdsourcing instead of lab experiments

How can technological developments be used for novel incentive

mechanisms

Usage of Crowdsourcing

How can we conduct reproducible experiments in diverse

crowdsourcing environment

How can we raise the awareness of crowdsourcing employers for the

individuals behind the crowd

Tobias Hoszligfeld

The crowd community in the lab

Tobias Hoszligfeld wwwmaswiwiuni-duede

bull Background Tobias Hoszligfeld

ndash 2004-2014 Research group ldquoFuture Internet Applicationsrdquo Chair of Communication Networks Univ of Wuumlrzburg

ndash Since Sep 2014 Chair of Modeling of Adaptive Systems (head) University of Duisburg-Essen

bull Interests Application of crowdsourcing

ndash User studies on Quality of Experience (QoE) for Internet apps eg video QoE in the crowd amp in the lab methodology reliability statisticshellip [2]

ndash Mobile crowdsourcing network and context measurements eg Wifi offloading incentives privacy recommendation mechanisms etc [1]

bull Interests The crowd community

ndash Crowd = Human beings Social network wisdom of the crowd community

ndash UnderstandingUtilizing the crowd community for successful human computation For which jobs can social interaction improve the quality of work Classification

ndash How can we deploy such a system where interaction is triggered

ndash System design Task design interaction mechanisms incentives

22 Nov 2015

[1] httpdropsdagstuhldeopusvolltexte20134354

[2] httpswww3informatikuni-wuerzburgdeqoewikiqualinetcrowd

Challenges Example of Crowdsourced QoE

Tobias Hoszligfeld wwwmaswiwiuni-duede

task design

incentives

Internet

crowd

community

anonymous

remote

subjects

Internet delivery of

test data and user

response

QoE study bull testing methodology bull QoE influence factors

under investigation bull additional QoE factors bull proper subset of items bull hellip

crowdsourcing setting bull short task duration bull instructions amp training bull feedback amp questions bull hidden influence factors bull reliability of users bull hellip

heterogeneous devices

and Internet access

implementation extrinsic vs intrinsic

motivation

Processing of data

Technical aspects bull server implementation bull client application amp

compatibility bull desired test conditions bull context monitoring bull hellip

test items

lab

scientific methodologies

22 Nov 2015

Jason Jacques

Jason Jacques supervised by Per Ola Kristensson

T

Crowdsourcing Code Perception

jtj21camacuk peopledscamacukjtj21

jtjacquesgmailcom jsonjcouk

PhD Working Title

science

Crowdsourcing Interests W

hat

W

ho

Ho

w

Wh

y jtj21camacuk jsonjcouk science

Radu Jianu

Radu Jianu Assistant Professor

Florida International University (PhD Brown University lsquo12)

Interests

Applications of eye-tracking in visualization tracking what people look at rather than where they look

bull Data intensive large scale eye-tracking experiments bull Gaze-contingent visualizations bull Gaze-driven data recommendation and big data curation

Automating user studies in data visualizations (on next page)

Automating user studies

1 Software designs that can (semi)automate a broad range of quantitative evaluations by leveraging

ndash Crowdsourcing ndash Benchmark data and task taxonomies ndash Automated result interpretation

2 Moving user evaluation from the end to within the software design and implementation process as a means to choose between competitive visual designs

Christian Keimel

copy IRT

Where

What

bull Hybrid broadcast broadband technologies

bull Big Data in broadcasting environment and content-creation

bull Audio-visual QoE assessment (with crowdsourcing)

About me

Christian Keimel Slide 60

Lecturer for Digital Broadcast Engineering

Research Engineer for Media Services and Platforms

copy IRT

Crowdsourcing in large scale optimisation

The human component of Big Data

Moving crowdsourcing to lean-back devices

Crowdsourcing from the couch

Crowdsourced experiments emulating the laboratory

Is it time for new methodologies ignoring the lab

Questionshellip

Christian Keimel Slide 61

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 35: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Sara Fabrikant

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

humanmdashsystem (visuo-spatial) interface

spatio-temporal analytics

spatialization human navigation

multivariate spatial analysis etc)

interface design

large amp small interactive (map) displays

(ie desktop mobile VR etc)

empirical evaluations

(ie eye tracking EDA EEG etc) [httpwwwunitsevenconz]

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

human sensing observatory set up

controlled laboratory | in-situ (messy) real world setting

semi|automatic data collection of various human behavioral data streams

ie psycho-physiological perceptual cognitive and sensory-motor behaviors (through crowdsourcing )

Brian Fisher

My double research life

Brian Fisher

bull PhD in Experimental Psych

bull Postdoc w Cognitive Science

society founder amp president

bull Psychonomics Fellow

bull VIS-related symposia at Cogsci

amp APS papers on cogsci of

interaction

bull Fuzzy-logicBayes models

bull Postdoc funded by Inst for Robotics

and AI

bull VAST SC VEC VACCINE

bull Co-organized Dagstuhl ldquoInteraction

with Information for Visual Reasoningrdquo

Cogsci-based papers at VIS CHI

BELIV

My Plan

VIS

Me Psychological Science

Ujwal Gadiraju

p = 0055111

Crowd Worker

Behavior

Microtask

Design

Crowdsourcing Paradigm Quality amp Effectiveness

UJWAL GADIRAJU

Incentivization Gamification

Neha Gupta

Getting to know the crowd

Presented at Dagstuhl by

Dr David Martin amp Neha Gupta

Neha Gupta

PhD Student School of Computer Science University of Nottingham UK

Investigating crowdwork through the platform Amazon Mechanical Turk

psxng1nottinghamacuk

Neha_Gi_ji

Research amp Research teams

Research Studies

bull Qualitative study of Turkers on online forums like mturkforum and Turker Nation by Dr David Martin XRCE

bull Ethnographic Studies of Turkers in their respective homes and places of work in India by Neha Gupta UoN

Research teams

Xerox

bull David Martin Jacki OrsquoNeill (Now MSR India) Ben Hanrahan (Now PSU)

psxng1nottinghamacuk

Neha_Gi_ji

Nottingham

bull Andy Crabtree

Tom Rodden

Matthias Hirth

Matthias Hirth 48 48

About Myself

Matthias Hirth

Research Assistant at University of Wuumlrzburg

httpmatthias-hirthcom

matthiashirthinformatikuni-wuerzburgde

Crowdsourcing

Platform analysis

Platform optimization

Human factors

Use cases

Online Social Networks

Collaboration networks

Structural differences of OSNs

Application of OSN relationships

Interactive applicationsGaming

Challenges of real time interactions

Serious gamingGamification

E-Sports broadcastinglive streaming

Quality of Experience

Monitoring techniques

Quantification techniques

(Crowd-bases) test design

Social interactions

Test methodology Incentive design

Use case

Measurement technique

Identification of

qualification

Matthias Hirth 49 49

Improving Crowdsourcing Technology and the Way We Use it

Crowdsourcing technology

How to transfer well known lab procedures into an uncontrolled

crowdsourcing environment

Which crowdsourcing platform features are required to conduct

successful experiments but are not present yet

Which technical developments (can) foster the usage of

crowdsourcing instead of lab experiments

How can technological developments be used for novel incentive

mechanisms

Usage of Crowdsourcing

How can we conduct reproducible experiments in diverse

crowdsourcing environment

How can we raise the awareness of crowdsourcing employers for the

individuals behind the crowd

Tobias Hoszligfeld

The crowd community in the lab

Tobias Hoszligfeld wwwmaswiwiuni-duede

bull Background Tobias Hoszligfeld

ndash 2004-2014 Research group ldquoFuture Internet Applicationsrdquo Chair of Communication Networks Univ of Wuumlrzburg

ndash Since Sep 2014 Chair of Modeling of Adaptive Systems (head) University of Duisburg-Essen

bull Interests Application of crowdsourcing

ndash User studies on Quality of Experience (QoE) for Internet apps eg video QoE in the crowd amp in the lab methodology reliability statisticshellip [2]

ndash Mobile crowdsourcing network and context measurements eg Wifi offloading incentives privacy recommendation mechanisms etc [1]

bull Interests The crowd community

ndash Crowd = Human beings Social network wisdom of the crowd community

ndash UnderstandingUtilizing the crowd community for successful human computation For which jobs can social interaction improve the quality of work Classification

ndash How can we deploy such a system where interaction is triggered

ndash System design Task design interaction mechanisms incentives

22 Nov 2015

[1] httpdropsdagstuhldeopusvolltexte20134354

[2] httpswww3informatikuni-wuerzburgdeqoewikiqualinetcrowd

Challenges Example of Crowdsourced QoE

Tobias Hoszligfeld wwwmaswiwiuni-duede

task design

incentives

Internet

crowd

community

anonymous

remote

subjects

Internet delivery of

test data and user

response

QoE study bull testing methodology bull QoE influence factors

under investigation bull additional QoE factors bull proper subset of items bull hellip

crowdsourcing setting bull short task duration bull instructions amp training bull feedback amp questions bull hidden influence factors bull reliability of users bull hellip

heterogeneous devices

and Internet access

implementation extrinsic vs intrinsic

motivation

Processing of data

Technical aspects bull server implementation bull client application amp

compatibility bull desired test conditions bull context monitoring bull hellip

test items

lab

scientific methodologies

22 Nov 2015

Jason Jacques

Jason Jacques supervised by Per Ola Kristensson

T

Crowdsourcing Code Perception

jtj21camacuk peopledscamacukjtj21

jtjacquesgmailcom jsonjcouk

PhD Working Title

science

Crowdsourcing Interests W

hat

W

ho

Ho

w

Wh

y jtj21camacuk jsonjcouk science

Radu Jianu

Radu Jianu Assistant Professor

Florida International University (PhD Brown University lsquo12)

Interests

Applications of eye-tracking in visualization tracking what people look at rather than where they look

bull Data intensive large scale eye-tracking experiments bull Gaze-contingent visualizations bull Gaze-driven data recommendation and big data curation

Automating user studies in data visualizations (on next page)

Automating user studies

1 Software designs that can (semi)automate a broad range of quantitative evaluations by leveraging

ndash Crowdsourcing ndash Benchmark data and task taxonomies ndash Automated result interpretation

2 Moving user evaluation from the end to within the software design and implementation process as a means to choose between competitive visual designs

Christian Keimel

copy IRT

Where

What

bull Hybrid broadcast broadband technologies

bull Big Data in broadcasting environment and content-creation

bull Audio-visual QoE assessment (with crowdsourcing)

About me

Christian Keimel Slide 60

Lecturer for Digital Broadcast Engineering

Research Engineer for Media Services and Platforms

copy IRT

Crowdsourcing in large scale optimisation

The human component of Big Data

Moving crowdsourcing to lean-back devices

Crowdsourcing from the couch

Crowdsourced experiments emulating the laboratory

Is it time for new methodologies ignoring the lab

Questionshellip

Christian Keimel Slide 61

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 36: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

humanmdashsystem (visuo-spatial) interface

spatio-temporal analytics

spatialization human navigation

multivariate spatial analysis etc)

interface design

large amp small interactive (map) displays

(ie desktop mobile VR etc)

empirical evaluations

(ie eye tracking EDA EEG etc) [httpwwwunitsevenconz]

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

human sensing observatory set up

controlled laboratory | in-situ (messy) real world setting

semi|automatic data collection of various human behavioral data streams

ie psycho-physiological perceptual cognitive and sensory-motor behaviors (through crowdsourcing )

Brian Fisher

My double research life

Brian Fisher

bull PhD in Experimental Psych

bull Postdoc w Cognitive Science

society founder amp president

bull Psychonomics Fellow

bull VIS-related symposia at Cogsci

amp APS papers on cogsci of

interaction

bull Fuzzy-logicBayes models

bull Postdoc funded by Inst for Robotics

and AI

bull VAST SC VEC VACCINE

bull Co-organized Dagstuhl ldquoInteraction

with Information for Visual Reasoningrdquo

Cogsci-based papers at VIS CHI

BELIV

My Plan

VIS

Me Psychological Science

Ujwal Gadiraju

p = 0055111

Crowd Worker

Behavior

Microtask

Design

Crowdsourcing Paradigm Quality amp Effectiveness

UJWAL GADIRAJU

Incentivization Gamification

Neha Gupta

Getting to know the crowd

Presented at Dagstuhl by

Dr David Martin amp Neha Gupta

Neha Gupta

PhD Student School of Computer Science University of Nottingham UK

Investigating crowdwork through the platform Amazon Mechanical Turk

psxng1nottinghamacuk

Neha_Gi_ji

Research amp Research teams

Research Studies

bull Qualitative study of Turkers on online forums like mturkforum and Turker Nation by Dr David Martin XRCE

bull Ethnographic Studies of Turkers in their respective homes and places of work in India by Neha Gupta UoN

Research teams

Xerox

bull David Martin Jacki OrsquoNeill (Now MSR India) Ben Hanrahan (Now PSU)

psxng1nottinghamacuk

Neha_Gi_ji

Nottingham

bull Andy Crabtree

Tom Rodden

Matthias Hirth

Matthias Hirth 48 48

About Myself

Matthias Hirth

Research Assistant at University of Wuumlrzburg

httpmatthias-hirthcom

matthiashirthinformatikuni-wuerzburgde

Crowdsourcing

Platform analysis

Platform optimization

Human factors

Use cases

Online Social Networks

Collaboration networks

Structural differences of OSNs

Application of OSN relationships

Interactive applicationsGaming

Challenges of real time interactions

Serious gamingGamification

E-Sports broadcastinglive streaming

Quality of Experience

Monitoring techniques

Quantification techniques

(Crowd-bases) test design

Social interactions

Test methodology Incentive design

Use case

Measurement technique

Identification of

qualification

Matthias Hirth 49 49

Improving Crowdsourcing Technology and the Way We Use it

Crowdsourcing technology

How to transfer well known lab procedures into an uncontrolled

crowdsourcing environment

Which crowdsourcing platform features are required to conduct

successful experiments but are not present yet

Which technical developments (can) foster the usage of

crowdsourcing instead of lab experiments

How can technological developments be used for novel incentive

mechanisms

Usage of Crowdsourcing

How can we conduct reproducible experiments in diverse

crowdsourcing environment

How can we raise the awareness of crowdsourcing employers for the

individuals behind the crowd

Tobias Hoszligfeld

The crowd community in the lab

Tobias Hoszligfeld wwwmaswiwiuni-duede

bull Background Tobias Hoszligfeld

ndash 2004-2014 Research group ldquoFuture Internet Applicationsrdquo Chair of Communication Networks Univ of Wuumlrzburg

ndash Since Sep 2014 Chair of Modeling of Adaptive Systems (head) University of Duisburg-Essen

bull Interests Application of crowdsourcing

ndash User studies on Quality of Experience (QoE) for Internet apps eg video QoE in the crowd amp in the lab methodology reliability statisticshellip [2]

ndash Mobile crowdsourcing network and context measurements eg Wifi offloading incentives privacy recommendation mechanisms etc [1]

bull Interests The crowd community

ndash Crowd = Human beings Social network wisdom of the crowd community

ndash UnderstandingUtilizing the crowd community for successful human computation For which jobs can social interaction improve the quality of work Classification

ndash How can we deploy such a system where interaction is triggered

ndash System design Task design interaction mechanisms incentives

22 Nov 2015

[1] httpdropsdagstuhldeopusvolltexte20134354

[2] httpswww3informatikuni-wuerzburgdeqoewikiqualinetcrowd

Challenges Example of Crowdsourced QoE

Tobias Hoszligfeld wwwmaswiwiuni-duede

task design

incentives

Internet

crowd

community

anonymous

remote

subjects

Internet delivery of

test data and user

response

QoE study bull testing methodology bull QoE influence factors

under investigation bull additional QoE factors bull proper subset of items bull hellip

crowdsourcing setting bull short task duration bull instructions amp training bull feedback amp questions bull hidden influence factors bull reliability of users bull hellip

heterogeneous devices

and Internet access

implementation extrinsic vs intrinsic

motivation

Processing of data

Technical aspects bull server implementation bull client application amp

compatibility bull desired test conditions bull context monitoring bull hellip

test items

lab

scientific methodologies

22 Nov 2015

Jason Jacques

Jason Jacques supervised by Per Ola Kristensson

T

Crowdsourcing Code Perception

jtj21camacuk peopledscamacukjtj21

jtjacquesgmailcom jsonjcouk

PhD Working Title

science

Crowdsourcing Interests W

hat

W

ho

Ho

w

Wh

y jtj21camacuk jsonjcouk science

Radu Jianu

Radu Jianu Assistant Professor

Florida International University (PhD Brown University lsquo12)

Interests

Applications of eye-tracking in visualization tracking what people look at rather than where they look

bull Data intensive large scale eye-tracking experiments bull Gaze-contingent visualizations bull Gaze-driven data recommendation and big data curation

Automating user studies in data visualizations (on next page)

Automating user studies

1 Software designs that can (semi)automate a broad range of quantitative evaluations by leveraging

ndash Crowdsourcing ndash Benchmark data and task taxonomies ndash Automated result interpretation

2 Moving user evaluation from the end to within the software design and implementation process as a means to choose between competitive visual designs

Christian Keimel

copy IRT

Where

What

bull Hybrid broadcast broadband technologies

bull Big Data in broadcasting environment and content-creation

bull Audio-visual QoE assessment (with crowdsourcing)

About me

Christian Keimel Slide 60

Lecturer for Digital Broadcast Engineering

Research Engineer for Media Services and Platforms

copy IRT

Crowdsourcing in large scale optimisation

The human component of Big Data

Moving crowdsourcing to lean-back devices

Crowdsourcing from the couch

Crowdsourced experiments emulating the laboratory

Is it time for new methodologies ignoring the lab

Questionshellip

Christian Keimel Slide 61

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 37: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

s a r a irina fabrikant geographer | giscientist | cognitive mapepatician

human sensing observatory set up

controlled laboratory | in-situ (messy) real world setting

semi|automatic data collection of various human behavioral data streams

ie psycho-physiological perceptual cognitive and sensory-motor behaviors (through crowdsourcing )

Brian Fisher

My double research life

Brian Fisher

bull PhD in Experimental Psych

bull Postdoc w Cognitive Science

society founder amp president

bull Psychonomics Fellow

bull VIS-related symposia at Cogsci

amp APS papers on cogsci of

interaction

bull Fuzzy-logicBayes models

bull Postdoc funded by Inst for Robotics

and AI

bull VAST SC VEC VACCINE

bull Co-organized Dagstuhl ldquoInteraction

with Information for Visual Reasoningrdquo

Cogsci-based papers at VIS CHI

BELIV

My Plan

VIS

Me Psychological Science

Ujwal Gadiraju

p = 0055111

Crowd Worker

Behavior

Microtask

Design

Crowdsourcing Paradigm Quality amp Effectiveness

UJWAL GADIRAJU

Incentivization Gamification

Neha Gupta

Getting to know the crowd

Presented at Dagstuhl by

Dr David Martin amp Neha Gupta

Neha Gupta

PhD Student School of Computer Science University of Nottingham UK

Investigating crowdwork through the platform Amazon Mechanical Turk

psxng1nottinghamacuk

Neha_Gi_ji

Research amp Research teams

Research Studies

bull Qualitative study of Turkers on online forums like mturkforum and Turker Nation by Dr David Martin XRCE

bull Ethnographic Studies of Turkers in their respective homes and places of work in India by Neha Gupta UoN

Research teams

Xerox

bull David Martin Jacki OrsquoNeill (Now MSR India) Ben Hanrahan (Now PSU)

psxng1nottinghamacuk

Neha_Gi_ji

Nottingham

bull Andy Crabtree

Tom Rodden

Matthias Hirth

Matthias Hirth 48 48

About Myself

Matthias Hirth

Research Assistant at University of Wuumlrzburg

httpmatthias-hirthcom

matthiashirthinformatikuni-wuerzburgde

Crowdsourcing

Platform analysis

Platform optimization

Human factors

Use cases

Online Social Networks

Collaboration networks

Structural differences of OSNs

Application of OSN relationships

Interactive applicationsGaming

Challenges of real time interactions

Serious gamingGamification

E-Sports broadcastinglive streaming

Quality of Experience

Monitoring techniques

Quantification techniques

(Crowd-bases) test design

Social interactions

Test methodology Incentive design

Use case

Measurement technique

Identification of

qualification

Matthias Hirth 49 49

Improving Crowdsourcing Technology and the Way We Use it

Crowdsourcing technology

How to transfer well known lab procedures into an uncontrolled

crowdsourcing environment

Which crowdsourcing platform features are required to conduct

successful experiments but are not present yet

Which technical developments (can) foster the usage of

crowdsourcing instead of lab experiments

How can technological developments be used for novel incentive

mechanisms

Usage of Crowdsourcing

How can we conduct reproducible experiments in diverse

crowdsourcing environment

How can we raise the awareness of crowdsourcing employers for the

individuals behind the crowd

Tobias Hoszligfeld

The crowd community in the lab

Tobias Hoszligfeld wwwmaswiwiuni-duede

bull Background Tobias Hoszligfeld

ndash 2004-2014 Research group ldquoFuture Internet Applicationsrdquo Chair of Communication Networks Univ of Wuumlrzburg

ndash Since Sep 2014 Chair of Modeling of Adaptive Systems (head) University of Duisburg-Essen

bull Interests Application of crowdsourcing

ndash User studies on Quality of Experience (QoE) for Internet apps eg video QoE in the crowd amp in the lab methodology reliability statisticshellip [2]

ndash Mobile crowdsourcing network and context measurements eg Wifi offloading incentives privacy recommendation mechanisms etc [1]

bull Interests The crowd community

ndash Crowd = Human beings Social network wisdom of the crowd community

ndash UnderstandingUtilizing the crowd community for successful human computation For which jobs can social interaction improve the quality of work Classification

ndash How can we deploy such a system where interaction is triggered

ndash System design Task design interaction mechanisms incentives

22 Nov 2015

[1] httpdropsdagstuhldeopusvolltexte20134354

[2] httpswww3informatikuni-wuerzburgdeqoewikiqualinetcrowd

Challenges Example of Crowdsourced QoE

Tobias Hoszligfeld wwwmaswiwiuni-duede

task design

incentives

Internet

crowd

community

anonymous

remote

subjects

Internet delivery of

test data and user

response

QoE study bull testing methodology bull QoE influence factors

under investigation bull additional QoE factors bull proper subset of items bull hellip

crowdsourcing setting bull short task duration bull instructions amp training bull feedback amp questions bull hidden influence factors bull reliability of users bull hellip

heterogeneous devices

and Internet access

implementation extrinsic vs intrinsic

motivation

Processing of data

Technical aspects bull server implementation bull client application amp

compatibility bull desired test conditions bull context monitoring bull hellip

test items

lab

scientific methodologies

22 Nov 2015

Jason Jacques

Jason Jacques supervised by Per Ola Kristensson

T

Crowdsourcing Code Perception

jtj21camacuk peopledscamacukjtj21

jtjacquesgmailcom jsonjcouk

PhD Working Title

science

Crowdsourcing Interests W

hat

W

ho

Ho

w

Wh

y jtj21camacuk jsonjcouk science

Radu Jianu

Radu Jianu Assistant Professor

Florida International University (PhD Brown University lsquo12)

Interests

Applications of eye-tracking in visualization tracking what people look at rather than where they look

bull Data intensive large scale eye-tracking experiments bull Gaze-contingent visualizations bull Gaze-driven data recommendation and big data curation

Automating user studies in data visualizations (on next page)

Automating user studies

1 Software designs that can (semi)automate a broad range of quantitative evaluations by leveraging

ndash Crowdsourcing ndash Benchmark data and task taxonomies ndash Automated result interpretation

2 Moving user evaluation from the end to within the software design and implementation process as a means to choose between competitive visual designs

Christian Keimel

copy IRT

Where

What

bull Hybrid broadcast broadband technologies

bull Big Data in broadcasting environment and content-creation

bull Audio-visual QoE assessment (with crowdsourcing)

About me

Christian Keimel Slide 60

Lecturer for Digital Broadcast Engineering

Research Engineer for Media Services and Platforms

copy IRT

Crowdsourcing in large scale optimisation

The human component of Big Data

Moving crowdsourcing to lean-back devices

Crowdsourcing from the couch

Crowdsourced experiments emulating the laboratory

Is it time for new methodologies ignoring the lab

Questionshellip

Christian Keimel Slide 61

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 38: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Brian Fisher

My double research life

Brian Fisher

bull PhD in Experimental Psych

bull Postdoc w Cognitive Science

society founder amp president

bull Psychonomics Fellow

bull VIS-related symposia at Cogsci

amp APS papers on cogsci of

interaction

bull Fuzzy-logicBayes models

bull Postdoc funded by Inst for Robotics

and AI

bull VAST SC VEC VACCINE

bull Co-organized Dagstuhl ldquoInteraction

with Information for Visual Reasoningrdquo

Cogsci-based papers at VIS CHI

BELIV

My Plan

VIS

Me Psychological Science

Ujwal Gadiraju

p = 0055111

Crowd Worker

Behavior

Microtask

Design

Crowdsourcing Paradigm Quality amp Effectiveness

UJWAL GADIRAJU

Incentivization Gamification

Neha Gupta

Getting to know the crowd

Presented at Dagstuhl by

Dr David Martin amp Neha Gupta

Neha Gupta

PhD Student School of Computer Science University of Nottingham UK

Investigating crowdwork through the platform Amazon Mechanical Turk

psxng1nottinghamacuk

Neha_Gi_ji

Research amp Research teams

Research Studies

bull Qualitative study of Turkers on online forums like mturkforum and Turker Nation by Dr David Martin XRCE

bull Ethnographic Studies of Turkers in their respective homes and places of work in India by Neha Gupta UoN

Research teams

Xerox

bull David Martin Jacki OrsquoNeill (Now MSR India) Ben Hanrahan (Now PSU)

psxng1nottinghamacuk

Neha_Gi_ji

Nottingham

bull Andy Crabtree

Tom Rodden

Matthias Hirth

Matthias Hirth 48 48

About Myself

Matthias Hirth

Research Assistant at University of Wuumlrzburg

httpmatthias-hirthcom

matthiashirthinformatikuni-wuerzburgde

Crowdsourcing

Platform analysis

Platform optimization

Human factors

Use cases

Online Social Networks

Collaboration networks

Structural differences of OSNs

Application of OSN relationships

Interactive applicationsGaming

Challenges of real time interactions

Serious gamingGamification

E-Sports broadcastinglive streaming

Quality of Experience

Monitoring techniques

Quantification techniques

(Crowd-bases) test design

Social interactions

Test methodology Incentive design

Use case

Measurement technique

Identification of

qualification

Matthias Hirth 49 49

Improving Crowdsourcing Technology and the Way We Use it

Crowdsourcing technology

How to transfer well known lab procedures into an uncontrolled

crowdsourcing environment

Which crowdsourcing platform features are required to conduct

successful experiments but are not present yet

Which technical developments (can) foster the usage of

crowdsourcing instead of lab experiments

How can technological developments be used for novel incentive

mechanisms

Usage of Crowdsourcing

How can we conduct reproducible experiments in diverse

crowdsourcing environment

How can we raise the awareness of crowdsourcing employers for the

individuals behind the crowd

Tobias Hoszligfeld

The crowd community in the lab

Tobias Hoszligfeld wwwmaswiwiuni-duede

bull Background Tobias Hoszligfeld

ndash 2004-2014 Research group ldquoFuture Internet Applicationsrdquo Chair of Communication Networks Univ of Wuumlrzburg

ndash Since Sep 2014 Chair of Modeling of Adaptive Systems (head) University of Duisburg-Essen

bull Interests Application of crowdsourcing

ndash User studies on Quality of Experience (QoE) for Internet apps eg video QoE in the crowd amp in the lab methodology reliability statisticshellip [2]

ndash Mobile crowdsourcing network and context measurements eg Wifi offloading incentives privacy recommendation mechanisms etc [1]

bull Interests The crowd community

ndash Crowd = Human beings Social network wisdom of the crowd community

ndash UnderstandingUtilizing the crowd community for successful human computation For which jobs can social interaction improve the quality of work Classification

ndash How can we deploy such a system where interaction is triggered

ndash System design Task design interaction mechanisms incentives

22 Nov 2015

[1] httpdropsdagstuhldeopusvolltexte20134354

[2] httpswww3informatikuni-wuerzburgdeqoewikiqualinetcrowd

Challenges Example of Crowdsourced QoE

Tobias Hoszligfeld wwwmaswiwiuni-duede

task design

incentives

Internet

crowd

community

anonymous

remote

subjects

Internet delivery of

test data and user

response

QoE study bull testing methodology bull QoE influence factors

under investigation bull additional QoE factors bull proper subset of items bull hellip

crowdsourcing setting bull short task duration bull instructions amp training bull feedback amp questions bull hidden influence factors bull reliability of users bull hellip

heterogeneous devices

and Internet access

implementation extrinsic vs intrinsic

motivation

Processing of data

Technical aspects bull server implementation bull client application amp

compatibility bull desired test conditions bull context monitoring bull hellip

test items

lab

scientific methodologies

22 Nov 2015

Jason Jacques

Jason Jacques supervised by Per Ola Kristensson

T

Crowdsourcing Code Perception

jtj21camacuk peopledscamacukjtj21

jtjacquesgmailcom jsonjcouk

PhD Working Title

science

Crowdsourcing Interests W

hat

W

ho

Ho

w

Wh

y jtj21camacuk jsonjcouk science

Radu Jianu

Radu Jianu Assistant Professor

Florida International University (PhD Brown University lsquo12)

Interests

Applications of eye-tracking in visualization tracking what people look at rather than where they look

bull Data intensive large scale eye-tracking experiments bull Gaze-contingent visualizations bull Gaze-driven data recommendation and big data curation

Automating user studies in data visualizations (on next page)

Automating user studies

1 Software designs that can (semi)automate a broad range of quantitative evaluations by leveraging

ndash Crowdsourcing ndash Benchmark data and task taxonomies ndash Automated result interpretation

2 Moving user evaluation from the end to within the software design and implementation process as a means to choose between competitive visual designs

Christian Keimel

copy IRT

Where

What

bull Hybrid broadcast broadband technologies

bull Big Data in broadcasting environment and content-creation

bull Audio-visual QoE assessment (with crowdsourcing)

About me

Christian Keimel Slide 60

Lecturer for Digital Broadcast Engineering

Research Engineer for Media Services and Platforms

copy IRT

Crowdsourcing in large scale optimisation

The human component of Big Data

Moving crowdsourcing to lean-back devices

Crowdsourcing from the couch

Crowdsourced experiments emulating the laboratory

Is it time for new methodologies ignoring the lab

Questionshellip

Christian Keimel Slide 61

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 39: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

My double research life

Brian Fisher

bull PhD in Experimental Psych

bull Postdoc w Cognitive Science

society founder amp president

bull Psychonomics Fellow

bull VIS-related symposia at Cogsci

amp APS papers on cogsci of

interaction

bull Fuzzy-logicBayes models

bull Postdoc funded by Inst for Robotics

and AI

bull VAST SC VEC VACCINE

bull Co-organized Dagstuhl ldquoInteraction

with Information for Visual Reasoningrdquo

Cogsci-based papers at VIS CHI

BELIV

My Plan

VIS

Me Psychological Science

Ujwal Gadiraju

p = 0055111

Crowd Worker

Behavior

Microtask

Design

Crowdsourcing Paradigm Quality amp Effectiveness

UJWAL GADIRAJU

Incentivization Gamification

Neha Gupta

Getting to know the crowd

Presented at Dagstuhl by

Dr David Martin amp Neha Gupta

Neha Gupta

PhD Student School of Computer Science University of Nottingham UK

Investigating crowdwork through the platform Amazon Mechanical Turk

psxng1nottinghamacuk

Neha_Gi_ji

Research amp Research teams

Research Studies

bull Qualitative study of Turkers on online forums like mturkforum and Turker Nation by Dr David Martin XRCE

bull Ethnographic Studies of Turkers in their respective homes and places of work in India by Neha Gupta UoN

Research teams

Xerox

bull David Martin Jacki OrsquoNeill (Now MSR India) Ben Hanrahan (Now PSU)

psxng1nottinghamacuk

Neha_Gi_ji

Nottingham

bull Andy Crabtree

Tom Rodden

Matthias Hirth

Matthias Hirth 48 48

About Myself

Matthias Hirth

Research Assistant at University of Wuumlrzburg

httpmatthias-hirthcom

matthiashirthinformatikuni-wuerzburgde

Crowdsourcing

Platform analysis

Platform optimization

Human factors

Use cases

Online Social Networks

Collaboration networks

Structural differences of OSNs

Application of OSN relationships

Interactive applicationsGaming

Challenges of real time interactions

Serious gamingGamification

E-Sports broadcastinglive streaming

Quality of Experience

Monitoring techniques

Quantification techniques

(Crowd-bases) test design

Social interactions

Test methodology Incentive design

Use case

Measurement technique

Identification of

qualification

Matthias Hirth 49 49

Improving Crowdsourcing Technology and the Way We Use it

Crowdsourcing technology

How to transfer well known lab procedures into an uncontrolled

crowdsourcing environment

Which crowdsourcing platform features are required to conduct

successful experiments but are not present yet

Which technical developments (can) foster the usage of

crowdsourcing instead of lab experiments

How can technological developments be used for novel incentive

mechanisms

Usage of Crowdsourcing

How can we conduct reproducible experiments in diverse

crowdsourcing environment

How can we raise the awareness of crowdsourcing employers for the

individuals behind the crowd

Tobias Hoszligfeld

The crowd community in the lab

Tobias Hoszligfeld wwwmaswiwiuni-duede

bull Background Tobias Hoszligfeld

ndash 2004-2014 Research group ldquoFuture Internet Applicationsrdquo Chair of Communication Networks Univ of Wuumlrzburg

ndash Since Sep 2014 Chair of Modeling of Adaptive Systems (head) University of Duisburg-Essen

bull Interests Application of crowdsourcing

ndash User studies on Quality of Experience (QoE) for Internet apps eg video QoE in the crowd amp in the lab methodology reliability statisticshellip [2]

ndash Mobile crowdsourcing network and context measurements eg Wifi offloading incentives privacy recommendation mechanisms etc [1]

bull Interests The crowd community

ndash Crowd = Human beings Social network wisdom of the crowd community

ndash UnderstandingUtilizing the crowd community for successful human computation For which jobs can social interaction improve the quality of work Classification

ndash How can we deploy such a system where interaction is triggered

ndash System design Task design interaction mechanisms incentives

22 Nov 2015

[1] httpdropsdagstuhldeopusvolltexte20134354

[2] httpswww3informatikuni-wuerzburgdeqoewikiqualinetcrowd

Challenges Example of Crowdsourced QoE

Tobias Hoszligfeld wwwmaswiwiuni-duede

task design

incentives

Internet

crowd

community

anonymous

remote

subjects

Internet delivery of

test data and user

response

QoE study bull testing methodology bull QoE influence factors

under investigation bull additional QoE factors bull proper subset of items bull hellip

crowdsourcing setting bull short task duration bull instructions amp training bull feedback amp questions bull hidden influence factors bull reliability of users bull hellip

heterogeneous devices

and Internet access

implementation extrinsic vs intrinsic

motivation

Processing of data

Technical aspects bull server implementation bull client application amp

compatibility bull desired test conditions bull context monitoring bull hellip

test items

lab

scientific methodologies

22 Nov 2015

Jason Jacques

Jason Jacques supervised by Per Ola Kristensson

T

Crowdsourcing Code Perception

jtj21camacuk peopledscamacukjtj21

jtjacquesgmailcom jsonjcouk

PhD Working Title

science

Crowdsourcing Interests W

hat

W

ho

Ho

w

Wh

y jtj21camacuk jsonjcouk science

Radu Jianu

Radu Jianu Assistant Professor

Florida International University (PhD Brown University lsquo12)

Interests

Applications of eye-tracking in visualization tracking what people look at rather than where they look

bull Data intensive large scale eye-tracking experiments bull Gaze-contingent visualizations bull Gaze-driven data recommendation and big data curation

Automating user studies in data visualizations (on next page)

Automating user studies

1 Software designs that can (semi)automate a broad range of quantitative evaluations by leveraging

ndash Crowdsourcing ndash Benchmark data and task taxonomies ndash Automated result interpretation

2 Moving user evaluation from the end to within the software design and implementation process as a means to choose between competitive visual designs

Christian Keimel

copy IRT

Where

What

bull Hybrid broadcast broadband technologies

bull Big Data in broadcasting environment and content-creation

bull Audio-visual QoE assessment (with crowdsourcing)

About me

Christian Keimel Slide 60

Lecturer for Digital Broadcast Engineering

Research Engineer for Media Services and Platforms

copy IRT

Crowdsourcing in large scale optimisation

The human component of Big Data

Moving crowdsourcing to lean-back devices

Crowdsourcing from the couch

Crowdsourced experiments emulating the laboratory

Is it time for new methodologies ignoring the lab

Questionshellip

Christian Keimel Slide 61

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 40: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

My Plan

VIS

Me Psychological Science

Ujwal Gadiraju

p = 0055111

Crowd Worker

Behavior

Microtask

Design

Crowdsourcing Paradigm Quality amp Effectiveness

UJWAL GADIRAJU

Incentivization Gamification

Neha Gupta

Getting to know the crowd

Presented at Dagstuhl by

Dr David Martin amp Neha Gupta

Neha Gupta

PhD Student School of Computer Science University of Nottingham UK

Investigating crowdwork through the platform Amazon Mechanical Turk

psxng1nottinghamacuk

Neha_Gi_ji

Research amp Research teams

Research Studies

bull Qualitative study of Turkers on online forums like mturkforum and Turker Nation by Dr David Martin XRCE

bull Ethnographic Studies of Turkers in their respective homes and places of work in India by Neha Gupta UoN

Research teams

Xerox

bull David Martin Jacki OrsquoNeill (Now MSR India) Ben Hanrahan (Now PSU)

psxng1nottinghamacuk

Neha_Gi_ji

Nottingham

bull Andy Crabtree

Tom Rodden

Matthias Hirth

Matthias Hirth 48 48

About Myself

Matthias Hirth

Research Assistant at University of Wuumlrzburg

httpmatthias-hirthcom

matthiashirthinformatikuni-wuerzburgde

Crowdsourcing

Platform analysis

Platform optimization

Human factors

Use cases

Online Social Networks

Collaboration networks

Structural differences of OSNs

Application of OSN relationships

Interactive applicationsGaming

Challenges of real time interactions

Serious gamingGamification

E-Sports broadcastinglive streaming

Quality of Experience

Monitoring techniques

Quantification techniques

(Crowd-bases) test design

Social interactions

Test methodology Incentive design

Use case

Measurement technique

Identification of

qualification

Matthias Hirth 49 49

Improving Crowdsourcing Technology and the Way We Use it

Crowdsourcing technology

How to transfer well known lab procedures into an uncontrolled

crowdsourcing environment

Which crowdsourcing platform features are required to conduct

successful experiments but are not present yet

Which technical developments (can) foster the usage of

crowdsourcing instead of lab experiments

How can technological developments be used for novel incentive

mechanisms

Usage of Crowdsourcing

How can we conduct reproducible experiments in diverse

crowdsourcing environment

How can we raise the awareness of crowdsourcing employers for the

individuals behind the crowd

Tobias Hoszligfeld

The crowd community in the lab

Tobias Hoszligfeld wwwmaswiwiuni-duede

bull Background Tobias Hoszligfeld

ndash 2004-2014 Research group ldquoFuture Internet Applicationsrdquo Chair of Communication Networks Univ of Wuumlrzburg

ndash Since Sep 2014 Chair of Modeling of Adaptive Systems (head) University of Duisburg-Essen

bull Interests Application of crowdsourcing

ndash User studies on Quality of Experience (QoE) for Internet apps eg video QoE in the crowd amp in the lab methodology reliability statisticshellip [2]

ndash Mobile crowdsourcing network and context measurements eg Wifi offloading incentives privacy recommendation mechanisms etc [1]

bull Interests The crowd community

ndash Crowd = Human beings Social network wisdom of the crowd community

ndash UnderstandingUtilizing the crowd community for successful human computation For which jobs can social interaction improve the quality of work Classification

ndash How can we deploy such a system where interaction is triggered

ndash System design Task design interaction mechanisms incentives

22 Nov 2015

[1] httpdropsdagstuhldeopusvolltexte20134354

[2] httpswww3informatikuni-wuerzburgdeqoewikiqualinetcrowd

Challenges Example of Crowdsourced QoE

Tobias Hoszligfeld wwwmaswiwiuni-duede

task design

incentives

Internet

crowd

community

anonymous

remote

subjects

Internet delivery of

test data and user

response

QoE study bull testing methodology bull QoE influence factors

under investigation bull additional QoE factors bull proper subset of items bull hellip

crowdsourcing setting bull short task duration bull instructions amp training bull feedback amp questions bull hidden influence factors bull reliability of users bull hellip

heterogeneous devices

and Internet access

implementation extrinsic vs intrinsic

motivation

Processing of data

Technical aspects bull server implementation bull client application amp

compatibility bull desired test conditions bull context monitoring bull hellip

test items

lab

scientific methodologies

22 Nov 2015

Jason Jacques

Jason Jacques supervised by Per Ola Kristensson

T

Crowdsourcing Code Perception

jtj21camacuk peopledscamacukjtj21

jtjacquesgmailcom jsonjcouk

PhD Working Title

science

Crowdsourcing Interests W

hat

W

ho

Ho

w

Wh

y jtj21camacuk jsonjcouk science

Radu Jianu

Radu Jianu Assistant Professor

Florida International University (PhD Brown University lsquo12)

Interests

Applications of eye-tracking in visualization tracking what people look at rather than where they look

bull Data intensive large scale eye-tracking experiments bull Gaze-contingent visualizations bull Gaze-driven data recommendation and big data curation

Automating user studies in data visualizations (on next page)

Automating user studies

1 Software designs that can (semi)automate a broad range of quantitative evaluations by leveraging

ndash Crowdsourcing ndash Benchmark data and task taxonomies ndash Automated result interpretation

2 Moving user evaluation from the end to within the software design and implementation process as a means to choose between competitive visual designs

Christian Keimel

copy IRT

Where

What

bull Hybrid broadcast broadband technologies

bull Big Data in broadcasting environment and content-creation

bull Audio-visual QoE assessment (with crowdsourcing)

About me

Christian Keimel Slide 60

Lecturer for Digital Broadcast Engineering

Research Engineer for Media Services and Platforms

copy IRT

Crowdsourcing in large scale optimisation

The human component of Big Data

Moving crowdsourcing to lean-back devices

Crowdsourcing from the couch

Crowdsourced experiments emulating the laboratory

Is it time for new methodologies ignoring the lab

Questionshellip

Christian Keimel Slide 61

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 41: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Ujwal Gadiraju

p = 0055111

Crowd Worker

Behavior

Microtask

Design

Crowdsourcing Paradigm Quality amp Effectiveness

UJWAL GADIRAJU

Incentivization Gamification

Neha Gupta

Getting to know the crowd

Presented at Dagstuhl by

Dr David Martin amp Neha Gupta

Neha Gupta

PhD Student School of Computer Science University of Nottingham UK

Investigating crowdwork through the platform Amazon Mechanical Turk

psxng1nottinghamacuk

Neha_Gi_ji

Research amp Research teams

Research Studies

bull Qualitative study of Turkers on online forums like mturkforum and Turker Nation by Dr David Martin XRCE

bull Ethnographic Studies of Turkers in their respective homes and places of work in India by Neha Gupta UoN

Research teams

Xerox

bull David Martin Jacki OrsquoNeill (Now MSR India) Ben Hanrahan (Now PSU)

psxng1nottinghamacuk

Neha_Gi_ji

Nottingham

bull Andy Crabtree

Tom Rodden

Matthias Hirth

Matthias Hirth 48 48

About Myself

Matthias Hirth

Research Assistant at University of Wuumlrzburg

httpmatthias-hirthcom

matthiashirthinformatikuni-wuerzburgde

Crowdsourcing

Platform analysis

Platform optimization

Human factors

Use cases

Online Social Networks

Collaboration networks

Structural differences of OSNs

Application of OSN relationships

Interactive applicationsGaming

Challenges of real time interactions

Serious gamingGamification

E-Sports broadcastinglive streaming

Quality of Experience

Monitoring techniques

Quantification techniques

(Crowd-bases) test design

Social interactions

Test methodology Incentive design

Use case

Measurement technique

Identification of

qualification

Matthias Hirth 49 49

Improving Crowdsourcing Technology and the Way We Use it

Crowdsourcing technology

How to transfer well known lab procedures into an uncontrolled

crowdsourcing environment

Which crowdsourcing platform features are required to conduct

successful experiments but are not present yet

Which technical developments (can) foster the usage of

crowdsourcing instead of lab experiments

How can technological developments be used for novel incentive

mechanisms

Usage of Crowdsourcing

How can we conduct reproducible experiments in diverse

crowdsourcing environment

How can we raise the awareness of crowdsourcing employers for the

individuals behind the crowd

Tobias Hoszligfeld

The crowd community in the lab

Tobias Hoszligfeld wwwmaswiwiuni-duede

bull Background Tobias Hoszligfeld

ndash 2004-2014 Research group ldquoFuture Internet Applicationsrdquo Chair of Communication Networks Univ of Wuumlrzburg

ndash Since Sep 2014 Chair of Modeling of Adaptive Systems (head) University of Duisburg-Essen

bull Interests Application of crowdsourcing

ndash User studies on Quality of Experience (QoE) for Internet apps eg video QoE in the crowd amp in the lab methodology reliability statisticshellip [2]

ndash Mobile crowdsourcing network and context measurements eg Wifi offloading incentives privacy recommendation mechanisms etc [1]

bull Interests The crowd community

ndash Crowd = Human beings Social network wisdom of the crowd community

ndash UnderstandingUtilizing the crowd community for successful human computation For which jobs can social interaction improve the quality of work Classification

ndash How can we deploy such a system where interaction is triggered

ndash System design Task design interaction mechanisms incentives

22 Nov 2015

[1] httpdropsdagstuhldeopusvolltexte20134354

[2] httpswww3informatikuni-wuerzburgdeqoewikiqualinetcrowd

Challenges Example of Crowdsourced QoE

Tobias Hoszligfeld wwwmaswiwiuni-duede

task design

incentives

Internet

crowd

community

anonymous

remote

subjects

Internet delivery of

test data and user

response

QoE study bull testing methodology bull QoE influence factors

under investigation bull additional QoE factors bull proper subset of items bull hellip

crowdsourcing setting bull short task duration bull instructions amp training bull feedback amp questions bull hidden influence factors bull reliability of users bull hellip

heterogeneous devices

and Internet access

implementation extrinsic vs intrinsic

motivation

Processing of data

Technical aspects bull server implementation bull client application amp

compatibility bull desired test conditions bull context monitoring bull hellip

test items

lab

scientific methodologies

22 Nov 2015

Jason Jacques

Jason Jacques supervised by Per Ola Kristensson

T

Crowdsourcing Code Perception

jtj21camacuk peopledscamacukjtj21

jtjacquesgmailcom jsonjcouk

PhD Working Title

science

Crowdsourcing Interests W

hat

W

ho

Ho

w

Wh

y jtj21camacuk jsonjcouk science

Radu Jianu

Radu Jianu Assistant Professor

Florida International University (PhD Brown University lsquo12)

Interests

Applications of eye-tracking in visualization tracking what people look at rather than where they look

bull Data intensive large scale eye-tracking experiments bull Gaze-contingent visualizations bull Gaze-driven data recommendation and big data curation

Automating user studies in data visualizations (on next page)

Automating user studies

1 Software designs that can (semi)automate a broad range of quantitative evaluations by leveraging

ndash Crowdsourcing ndash Benchmark data and task taxonomies ndash Automated result interpretation

2 Moving user evaluation from the end to within the software design and implementation process as a means to choose between competitive visual designs

Christian Keimel

copy IRT

Where

What

bull Hybrid broadcast broadband technologies

bull Big Data in broadcasting environment and content-creation

bull Audio-visual QoE assessment (with crowdsourcing)

About me

Christian Keimel Slide 60

Lecturer for Digital Broadcast Engineering

Research Engineer for Media Services and Platforms

copy IRT

Crowdsourcing in large scale optimisation

The human component of Big Data

Moving crowdsourcing to lean-back devices

Crowdsourcing from the couch

Crowdsourced experiments emulating the laboratory

Is it time for new methodologies ignoring the lab

Questionshellip

Christian Keimel Slide 61

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 42: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

p = 0055111

Crowd Worker

Behavior

Microtask

Design

Crowdsourcing Paradigm Quality amp Effectiveness

UJWAL GADIRAJU

Incentivization Gamification

Neha Gupta

Getting to know the crowd

Presented at Dagstuhl by

Dr David Martin amp Neha Gupta

Neha Gupta

PhD Student School of Computer Science University of Nottingham UK

Investigating crowdwork through the platform Amazon Mechanical Turk

psxng1nottinghamacuk

Neha_Gi_ji

Research amp Research teams

Research Studies

bull Qualitative study of Turkers on online forums like mturkforum and Turker Nation by Dr David Martin XRCE

bull Ethnographic Studies of Turkers in their respective homes and places of work in India by Neha Gupta UoN

Research teams

Xerox

bull David Martin Jacki OrsquoNeill (Now MSR India) Ben Hanrahan (Now PSU)

psxng1nottinghamacuk

Neha_Gi_ji

Nottingham

bull Andy Crabtree

Tom Rodden

Matthias Hirth

Matthias Hirth 48 48

About Myself

Matthias Hirth

Research Assistant at University of Wuumlrzburg

httpmatthias-hirthcom

matthiashirthinformatikuni-wuerzburgde

Crowdsourcing

Platform analysis

Platform optimization

Human factors

Use cases

Online Social Networks

Collaboration networks

Structural differences of OSNs

Application of OSN relationships

Interactive applicationsGaming

Challenges of real time interactions

Serious gamingGamification

E-Sports broadcastinglive streaming

Quality of Experience

Monitoring techniques

Quantification techniques

(Crowd-bases) test design

Social interactions

Test methodology Incentive design

Use case

Measurement technique

Identification of

qualification

Matthias Hirth 49 49

Improving Crowdsourcing Technology and the Way We Use it

Crowdsourcing technology

How to transfer well known lab procedures into an uncontrolled

crowdsourcing environment

Which crowdsourcing platform features are required to conduct

successful experiments but are not present yet

Which technical developments (can) foster the usage of

crowdsourcing instead of lab experiments

How can technological developments be used for novel incentive

mechanisms

Usage of Crowdsourcing

How can we conduct reproducible experiments in diverse

crowdsourcing environment

How can we raise the awareness of crowdsourcing employers for the

individuals behind the crowd

Tobias Hoszligfeld

The crowd community in the lab

Tobias Hoszligfeld wwwmaswiwiuni-duede

bull Background Tobias Hoszligfeld

ndash 2004-2014 Research group ldquoFuture Internet Applicationsrdquo Chair of Communication Networks Univ of Wuumlrzburg

ndash Since Sep 2014 Chair of Modeling of Adaptive Systems (head) University of Duisburg-Essen

bull Interests Application of crowdsourcing

ndash User studies on Quality of Experience (QoE) for Internet apps eg video QoE in the crowd amp in the lab methodology reliability statisticshellip [2]

ndash Mobile crowdsourcing network and context measurements eg Wifi offloading incentives privacy recommendation mechanisms etc [1]

bull Interests The crowd community

ndash Crowd = Human beings Social network wisdom of the crowd community

ndash UnderstandingUtilizing the crowd community for successful human computation For which jobs can social interaction improve the quality of work Classification

ndash How can we deploy such a system where interaction is triggered

ndash System design Task design interaction mechanisms incentives

22 Nov 2015

[1] httpdropsdagstuhldeopusvolltexte20134354

[2] httpswww3informatikuni-wuerzburgdeqoewikiqualinetcrowd

Challenges Example of Crowdsourced QoE

Tobias Hoszligfeld wwwmaswiwiuni-duede

task design

incentives

Internet

crowd

community

anonymous

remote

subjects

Internet delivery of

test data and user

response

QoE study bull testing methodology bull QoE influence factors

under investigation bull additional QoE factors bull proper subset of items bull hellip

crowdsourcing setting bull short task duration bull instructions amp training bull feedback amp questions bull hidden influence factors bull reliability of users bull hellip

heterogeneous devices

and Internet access

implementation extrinsic vs intrinsic

motivation

Processing of data

Technical aspects bull server implementation bull client application amp

compatibility bull desired test conditions bull context monitoring bull hellip

test items

lab

scientific methodologies

22 Nov 2015

Jason Jacques

Jason Jacques supervised by Per Ola Kristensson

T

Crowdsourcing Code Perception

jtj21camacuk peopledscamacukjtj21

jtjacquesgmailcom jsonjcouk

PhD Working Title

science

Crowdsourcing Interests W

hat

W

ho

Ho

w

Wh

y jtj21camacuk jsonjcouk science

Radu Jianu

Radu Jianu Assistant Professor

Florida International University (PhD Brown University lsquo12)

Interests

Applications of eye-tracking in visualization tracking what people look at rather than where they look

bull Data intensive large scale eye-tracking experiments bull Gaze-contingent visualizations bull Gaze-driven data recommendation and big data curation

Automating user studies in data visualizations (on next page)

Automating user studies

1 Software designs that can (semi)automate a broad range of quantitative evaluations by leveraging

ndash Crowdsourcing ndash Benchmark data and task taxonomies ndash Automated result interpretation

2 Moving user evaluation from the end to within the software design and implementation process as a means to choose between competitive visual designs

Christian Keimel

copy IRT

Where

What

bull Hybrid broadcast broadband technologies

bull Big Data in broadcasting environment and content-creation

bull Audio-visual QoE assessment (with crowdsourcing)

About me

Christian Keimel Slide 60

Lecturer for Digital Broadcast Engineering

Research Engineer for Media Services and Platforms

copy IRT

Crowdsourcing in large scale optimisation

The human component of Big Data

Moving crowdsourcing to lean-back devices

Crowdsourcing from the couch

Crowdsourced experiments emulating the laboratory

Is it time for new methodologies ignoring the lab

Questionshellip

Christian Keimel Slide 61

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 43: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Crowd Worker

Behavior

Microtask

Design

Crowdsourcing Paradigm Quality amp Effectiveness

UJWAL GADIRAJU

Incentivization Gamification

Neha Gupta

Getting to know the crowd

Presented at Dagstuhl by

Dr David Martin amp Neha Gupta

Neha Gupta

PhD Student School of Computer Science University of Nottingham UK

Investigating crowdwork through the platform Amazon Mechanical Turk

psxng1nottinghamacuk

Neha_Gi_ji

Research amp Research teams

Research Studies

bull Qualitative study of Turkers on online forums like mturkforum and Turker Nation by Dr David Martin XRCE

bull Ethnographic Studies of Turkers in their respective homes and places of work in India by Neha Gupta UoN

Research teams

Xerox

bull David Martin Jacki OrsquoNeill (Now MSR India) Ben Hanrahan (Now PSU)

psxng1nottinghamacuk

Neha_Gi_ji

Nottingham

bull Andy Crabtree

Tom Rodden

Matthias Hirth

Matthias Hirth 48 48

About Myself

Matthias Hirth

Research Assistant at University of Wuumlrzburg

httpmatthias-hirthcom

matthiashirthinformatikuni-wuerzburgde

Crowdsourcing

Platform analysis

Platform optimization

Human factors

Use cases

Online Social Networks

Collaboration networks

Structural differences of OSNs

Application of OSN relationships

Interactive applicationsGaming

Challenges of real time interactions

Serious gamingGamification

E-Sports broadcastinglive streaming

Quality of Experience

Monitoring techniques

Quantification techniques

(Crowd-bases) test design

Social interactions

Test methodology Incentive design

Use case

Measurement technique

Identification of

qualification

Matthias Hirth 49 49

Improving Crowdsourcing Technology and the Way We Use it

Crowdsourcing technology

How to transfer well known lab procedures into an uncontrolled

crowdsourcing environment

Which crowdsourcing platform features are required to conduct

successful experiments but are not present yet

Which technical developments (can) foster the usage of

crowdsourcing instead of lab experiments

How can technological developments be used for novel incentive

mechanisms

Usage of Crowdsourcing

How can we conduct reproducible experiments in diverse

crowdsourcing environment

How can we raise the awareness of crowdsourcing employers for the

individuals behind the crowd

Tobias Hoszligfeld

The crowd community in the lab

Tobias Hoszligfeld wwwmaswiwiuni-duede

bull Background Tobias Hoszligfeld

ndash 2004-2014 Research group ldquoFuture Internet Applicationsrdquo Chair of Communication Networks Univ of Wuumlrzburg

ndash Since Sep 2014 Chair of Modeling of Adaptive Systems (head) University of Duisburg-Essen

bull Interests Application of crowdsourcing

ndash User studies on Quality of Experience (QoE) for Internet apps eg video QoE in the crowd amp in the lab methodology reliability statisticshellip [2]

ndash Mobile crowdsourcing network and context measurements eg Wifi offloading incentives privacy recommendation mechanisms etc [1]

bull Interests The crowd community

ndash Crowd = Human beings Social network wisdom of the crowd community

ndash UnderstandingUtilizing the crowd community for successful human computation For which jobs can social interaction improve the quality of work Classification

ndash How can we deploy such a system where interaction is triggered

ndash System design Task design interaction mechanisms incentives

22 Nov 2015

[1] httpdropsdagstuhldeopusvolltexte20134354

[2] httpswww3informatikuni-wuerzburgdeqoewikiqualinetcrowd

Challenges Example of Crowdsourced QoE

Tobias Hoszligfeld wwwmaswiwiuni-duede

task design

incentives

Internet

crowd

community

anonymous

remote

subjects

Internet delivery of

test data and user

response

QoE study bull testing methodology bull QoE influence factors

under investigation bull additional QoE factors bull proper subset of items bull hellip

crowdsourcing setting bull short task duration bull instructions amp training bull feedback amp questions bull hidden influence factors bull reliability of users bull hellip

heterogeneous devices

and Internet access

implementation extrinsic vs intrinsic

motivation

Processing of data

Technical aspects bull server implementation bull client application amp

compatibility bull desired test conditions bull context monitoring bull hellip

test items

lab

scientific methodologies

22 Nov 2015

Jason Jacques

Jason Jacques supervised by Per Ola Kristensson

T

Crowdsourcing Code Perception

jtj21camacuk peopledscamacukjtj21

jtjacquesgmailcom jsonjcouk

PhD Working Title

science

Crowdsourcing Interests W

hat

W

ho

Ho

w

Wh

y jtj21camacuk jsonjcouk science

Radu Jianu

Radu Jianu Assistant Professor

Florida International University (PhD Brown University lsquo12)

Interests

Applications of eye-tracking in visualization tracking what people look at rather than where they look

bull Data intensive large scale eye-tracking experiments bull Gaze-contingent visualizations bull Gaze-driven data recommendation and big data curation

Automating user studies in data visualizations (on next page)

Automating user studies

1 Software designs that can (semi)automate a broad range of quantitative evaluations by leveraging

ndash Crowdsourcing ndash Benchmark data and task taxonomies ndash Automated result interpretation

2 Moving user evaluation from the end to within the software design and implementation process as a means to choose between competitive visual designs

Christian Keimel

copy IRT

Where

What

bull Hybrid broadcast broadband technologies

bull Big Data in broadcasting environment and content-creation

bull Audio-visual QoE assessment (with crowdsourcing)

About me

Christian Keimel Slide 60

Lecturer for Digital Broadcast Engineering

Research Engineer for Media Services and Platforms

copy IRT

Crowdsourcing in large scale optimisation

The human component of Big Data

Moving crowdsourcing to lean-back devices

Crowdsourcing from the couch

Crowdsourced experiments emulating the laboratory

Is it time for new methodologies ignoring the lab

Questionshellip

Christian Keimel Slide 61

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 44: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Neha Gupta

Getting to know the crowd

Presented at Dagstuhl by

Dr David Martin amp Neha Gupta

Neha Gupta

PhD Student School of Computer Science University of Nottingham UK

Investigating crowdwork through the platform Amazon Mechanical Turk

psxng1nottinghamacuk

Neha_Gi_ji

Research amp Research teams

Research Studies

bull Qualitative study of Turkers on online forums like mturkforum and Turker Nation by Dr David Martin XRCE

bull Ethnographic Studies of Turkers in their respective homes and places of work in India by Neha Gupta UoN

Research teams

Xerox

bull David Martin Jacki OrsquoNeill (Now MSR India) Ben Hanrahan (Now PSU)

psxng1nottinghamacuk

Neha_Gi_ji

Nottingham

bull Andy Crabtree

Tom Rodden

Matthias Hirth

Matthias Hirth 48 48

About Myself

Matthias Hirth

Research Assistant at University of Wuumlrzburg

httpmatthias-hirthcom

matthiashirthinformatikuni-wuerzburgde

Crowdsourcing

Platform analysis

Platform optimization

Human factors

Use cases

Online Social Networks

Collaboration networks

Structural differences of OSNs

Application of OSN relationships

Interactive applicationsGaming

Challenges of real time interactions

Serious gamingGamification

E-Sports broadcastinglive streaming

Quality of Experience

Monitoring techniques

Quantification techniques

(Crowd-bases) test design

Social interactions

Test methodology Incentive design

Use case

Measurement technique

Identification of

qualification

Matthias Hirth 49 49

Improving Crowdsourcing Technology and the Way We Use it

Crowdsourcing technology

How to transfer well known lab procedures into an uncontrolled

crowdsourcing environment

Which crowdsourcing platform features are required to conduct

successful experiments but are not present yet

Which technical developments (can) foster the usage of

crowdsourcing instead of lab experiments

How can technological developments be used for novel incentive

mechanisms

Usage of Crowdsourcing

How can we conduct reproducible experiments in diverse

crowdsourcing environment

How can we raise the awareness of crowdsourcing employers for the

individuals behind the crowd

Tobias Hoszligfeld

The crowd community in the lab

Tobias Hoszligfeld wwwmaswiwiuni-duede

bull Background Tobias Hoszligfeld

ndash 2004-2014 Research group ldquoFuture Internet Applicationsrdquo Chair of Communication Networks Univ of Wuumlrzburg

ndash Since Sep 2014 Chair of Modeling of Adaptive Systems (head) University of Duisburg-Essen

bull Interests Application of crowdsourcing

ndash User studies on Quality of Experience (QoE) for Internet apps eg video QoE in the crowd amp in the lab methodology reliability statisticshellip [2]

ndash Mobile crowdsourcing network and context measurements eg Wifi offloading incentives privacy recommendation mechanisms etc [1]

bull Interests The crowd community

ndash Crowd = Human beings Social network wisdom of the crowd community

ndash UnderstandingUtilizing the crowd community for successful human computation For which jobs can social interaction improve the quality of work Classification

ndash How can we deploy such a system where interaction is triggered

ndash System design Task design interaction mechanisms incentives

22 Nov 2015

[1] httpdropsdagstuhldeopusvolltexte20134354

[2] httpswww3informatikuni-wuerzburgdeqoewikiqualinetcrowd

Challenges Example of Crowdsourced QoE

Tobias Hoszligfeld wwwmaswiwiuni-duede

task design

incentives

Internet

crowd

community

anonymous

remote

subjects

Internet delivery of

test data and user

response

QoE study bull testing methodology bull QoE influence factors

under investigation bull additional QoE factors bull proper subset of items bull hellip

crowdsourcing setting bull short task duration bull instructions amp training bull feedback amp questions bull hidden influence factors bull reliability of users bull hellip

heterogeneous devices

and Internet access

implementation extrinsic vs intrinsic

motivation

Processing of data

Technical aspects bull server implementation bull client application amp

compatibility bull desired test conditions bull context monitoring bull hellip

test items

lab

scientific methodologies

22 Nov 2015

Jason Jacques

Jason Jacques supervised by Per Ola Kristensson

T

Crowdsourcing Code Perception

jtj21camacuk peopledscamacukjtj21

jtjacquesgmailcom jsonjcouk

PhD Working Title

science

Crowdsourcing Interests W

hat

W

ho

Ho

w

Wh

y jtj21camacuk jsonjcouk science

Radu Jianu

Radu Jianu Assistant Professor

Florida International University (PhD Brown University lsquo12)

Interests

Applications of eye-tracking in visualization tracking what people look at rather than where they look

bull Data intensive large scale eye-tracking experiments bull Gaze-contingent visualizations bull Gaze-driven data recommendation and big data curation

Automating user studies in data visualizations (on next page)

Automating user studies

1 Software designs that can (semi)automate a broad range of quantitative evaluations by leveraging

ndash Crowdsourcing ndash Benchmark data and task taxonomies ndash Automated result interpretation

2 Moving user evaluation from the end to within the software design and implementation process as a means to choose between competitive visual designs

Christian Keimel

copy IRT

Where

What

bull Hybrid broadcast broadband technologies

bull Big Data in broadcasting environment and content-creation

bull Audio-visual QoE assessment (with crowdsourcing)

About me

Christian Keimel Slide 60

Lecturer for Digital Broadcast Engineering

Research Engineer for Media Services and Platforms

copy IRT

Crowdsourcing in large scale optimisation

The human component of Big Data

Moving crowdsourcing to lean-back devices

Crowdsourcing from the couch

Crowdsourced experiments emulating the laboratory

Is it time for new methodologies ignoring the lab

Questionshellip

Christian Keimel Slide 61

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 45: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Getting to know the crowd

Presented at Dagstuhl by

Dr David Martin amp Neha Gupta

Neha Gupta

PhD Student School of Computer Science University of Nottingham UK

Investigating crowdwork through the platform Amazon Mechanical Turk

psxng1nottinghamacuk

Neha_Gi_ji

Research amp Research teams

Research Studies

bull Qualitative study of Turkers on online forums like mturkforum and Turker Nation by Dr David Martin XRCE

bull Ethnographic Studies of Turkers in their respective homes and places of work in India by Neha Gupta UoN

Research teams

Xerox

bull David Martin Jacki OrsquoNeill (Now MSR India) Ben Hanrahan (Now PSU)

psxng1nottinghamacuk

Neha_Gi_ji

Nottingham

bull Andy Crabtree

Tom Rodden

Matthias Hirth

Matthias Hirth 48 48

About Myself

Matthias Hirth

Research Assistant at University of Wuumlrzburg

httpmatthias-hirthcom

matthiashirthinformatikuni-wuerzburgde

Crowdsourcing

Platform analysis

Platform optimization

Human factors

Use cases

Online Social Networks

Collaboration networks

Structural differences of OSNs

Application of OSN relationships

Interactive applicationsGaming

Challenges of real time interactions

Serious gamingGamification

E-Sports broadcastinglive streaming

Quality of Experience

Monitoring techniques

Quantification techniques

(Crowd-bases) test design

Social interactions

Test methodology Incentive design

Use case

Measurement technique

Identification of

qualification

Matthias Hirth 49 49

Improving Crowdsourcing Technology and the Way We Use it

Crowdsourcing technology

How to transfer well known lab procedures into an uncontrolled

crowdsourcing environment

Which crowdsourcing platform features are required to conduct

successful experiments but are not present yet

Which technical developments (can) foster the usage of

crowdsourcing instead of lab experiments

How can technological developments be used for novel incentive

mechanisms

Usage of Crowdsourcing

How can we conduct reproducible experiments in diverse

crowdsourcing environment

How can we raise the awareness of crowdsourcing employers for the

individuals behind the crowd

Tobias Hoszligfeld

The crowd community in the lab

Tobias Hoszligfeld wwwmaswiwiuni-duede

bull Background Tobias Hoszligfeld

ndash 2004-2014 Research group ldquoFuture Internet Applicationsrdquo Chair of Communication Networks Univ of Wuumlrzburg

ndash Since Sep 2014 Chair of Modeling of Adaptive Systems (head) University of Duisburg-Essen

bull Interests Application of crowdsourcing

ndash User studies on Quality of Experience (QoE) for Internet apps eg video QoE in the crowd amp in the lab methodology reliability statisticshellip [2]

ndash Mobile crowdsourcing network and context measurements eg Wifi offloading incentives privacy recommendation mechanisms etc [1]

bull Interests The crowd community

ndash Crowd = Human beings Social network wisdom of the crowd community

ndash UnderstandingUtilizing the crowd community for successful human computation For which jobs can social interaction improve the quality of work Classification

ndash How can we deploy such a system where interaction is triggered

ndash System design Task design interaction mechanisms incentives

22 Nov 2015

[1] httpdropsdagstuhldeopusvolltexte20134354

[2] httpswww3informatikuni-wuerzburgdeqoewikiqualinetcrowd

Challenges Example of Crowdsourced QoE

Tobias Hoszligfeld wwwmaswiwiuni-duede

task design

incentives

Internet

crowd

community

anonymous

remote

subjects

Internet delivery of

test data and user

response

QoE study bull testing methodology bull QoE influence factors

under investigation bull additional QoE factors bull proper subset of items bull hellip

crowdsourcing setting bull short task duration bull instructions amp training bull feedback amp questions bull hidden influence factors bull reliability of users bull hellip

heterogeneous devices

and Internet access

implementation extrinsic vs intrinsic

motivation

Processing of data

Technical aspects bull server implementation bull client application amp

compatibility bull desired test conditions bull context monitoring bull hellip

test items

lab

scientific methodologies

22 Nov 2015

Jason Jacques

Jason Jacques supervised by Per Ola Kristensson

T

Crowdsourcing Code Perception

jtj21camacuk peopledscamacukjtj21

jtjacquesgmailcom jsonjcouk

PhD Working Title

science

Crowdsourcing Interests W

hat

W

ho

Ho

w

Wh

y jtj21camacuk jsonjcouk science

Radu Jianu

Radu Jianu Assistant Professor

Florida International University (PhD Brown University lsquo12)

Interests

Applications of eye-tracking in visualization tracking what people look at rather than where they look

bull Data intensive large scale eye-tracking experiments bull Gaze-contingent visualizations bull Gaze-driven data recommendation and big data curation

Automating user studies in data visualizations (on next page)

Automating user studies

1 Software designs that can (semi)automate a broad range of quantitative evaluations by leveraging

ndash Crowdsourcing ndash Benchmark data and task taxonomies ndash Automated result interpretation

2 Moving user evaluation from the end to within the software design and implementation process as a means to choose between competitive visual designs

Christian Keimel

copy IRT

Where

What

bull Hybrid broadcast broadband technologies

bull Big Data in broadcasting environment and content-creation

bull Audio-visual QoE assessment (with crowdsourcing)

About me

Christian Keimel Slide 60

Lecturer for Digital Broadcast Engineering

Research Engineer for Media Services and Platforms

copy IRT

Crowdsourcing in large scale optimisation

The human component of Big Data

Moving crowdsourcing to lean-back devices

Crowdsourcing from the couch

Crowdsourced experiments emulating the laboratory

Is it time for new methodologies ignoring the lab

Questionshellip

Christian Keimel Slide 61

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 46: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Research amp Research teams

Research Studies

bull Qualitative study of Turkers on online forums like mturkforum and Turker Nation by Dr David Martin XRCE

bull Ethnographic Studies of Turkers in their respective homes and places of work in India by Neha Gupta UoN

Research teams

Xerox

bull David Martin Jacki OrsquoNeill (Now MSR India) Ben Hanrahan (Now PSU)

psxng1nottinghamacuk

Neha_Gi_ji

Nottingham

bull Andy Crabtree

Tom Rodden

Matthias Hirth

Matthias Hirth 48 48

About Myself

Matthias Hirth

Research Assistant at University of Wuumlrzburg

httpmatthias-hirthcom

matthiashirthinformatikuni-wuerzburgde

Crowdsourcing

Platform analysis

Platform optimization

Human factors

Use cases

Online Social Networks

Collaboration networks

Structural differences of OSNs

Application of OSN relationships

Interactive applicationsGaming

Challenges of real time interactions

Serious gamingGamification

E-Sports broadcastinglive streaming

Quality of Experience

Monitoring techniques

Quantification techniques

(Crowd-bases) test design

Social interactions

Test methodology Incentive design

Use case

Measurement technique

Identification of

qualification

Matthias Hirth 49 49

Improving Crowdsourcing Technology and the Way We Use it

Crowdsourcing technology

How to transfer well known lab procedures into an uncontrolled

crowdsourcing environment

Which crowdsourcing platform features are required to conduct

successful experiments but are not present yet

Which technical developments (can) foster the usage of

crowdsourcing instead of lab experiments

How can technological developments be used for novel incentive

mechanisms

Usage of Crowdsourcing

How can we conduct reproducible experiments in diverse

crowdsourcing environment

How can we raise the awareness of crowdsourcing employers for the

individuals behind the crowd

Tobias Hoszligfeld

The crowd community in the lab

Tobias Hoszligfeld wwwmaswiwiuni-duede

bull Background Tobias Hoszligfeld

ndash 2004-2014 Research group ldquoFuture Internet Applicationsrdquo Chair of Communication Networks Univ of Wuumlrzburg

ndash Since Sep 2014 Chair of Modeling of Adaptive Systems (head) University of Duisburg-Essen

bull Interests Application of crowdsourcing

ndash User studies on Quality of Experience (QoE) for Internet apps eg video QoE in the crowd amp in the lab methodology reliability statisticshellip [2]

ndash Mobile crowdsourcing network and context measurements eg Wifi offloading incentives privacy recommendation mechanisms etc [1]

bull Interests The crowd community

ndash Crowd = Human beings Social network wisdom of the crowd community

ndash UnderstandingUtilizing the crowd community for successful human computation For which jobs can social interaction improve the quality of work Classification

ndash How can we deploy such a system where interaction is triggered

ndash System design Task design interaction mechanisms incentives

22 Nov 2015

[1] httpdropsdagstuhldeopusvolltexte20134354

[2] httpswww3informatikuni-wuerzburgdeqoewikiqualinetcrowd

Challenges Example of Crowdsourced QoE

Tobias Hoszligfeld wwwmaswiwiuni-duede

task design

incentives

Internet

crowd

community

anonymous

remote

subjects

Internet delivery of

test data and user

response

QoE study bull testing methodology bull QoE influence factors

under investigation bull additional QoE factors bull proper subset of items bull hellip

crowdsourcing setting bull short task duration bull instructions amp training bull feedback amp questions bull hidden influence factors bull reliability of users bull hellip

heterogeneous devices

and Internet access

implementation extrinsic vs intrinsic

motivation

Processing of data

Technical aspects bull server implementation bull client application amp

compatibility bull desired test conditions bull context monitoring bull hellip

test items

lab

scientific methodologies

22 Nov 2015

Jason Jacques

Jason Jacques supervised by Per Ola Kristensson

T

Crowdsourcing Code Perception

jtj21camacuk peopledscamacukjtj21

jtjacquesgmailcom jsonjcouk

PhD Working Title

science

Crowdsourcing Interests W

hat

W

ho

Ho

w

Wh

y jtj21camacuk jsonjcouk science

Radu Jianu

Radu Jianu Assistant Professor

Florida International University (PhD Brown University lsquo12)

Interests

Applications of eye-tracking in visualization tracking what people look at rather than where they look

bull Data intensive large scale eye-tracking experiments bull Gaze-contingent visualizations bull Gaze-driven data recommendation and big data curation

Automating user studies in data visualizations (on next page)

Automating user studies

1 Software designs that can (semi)automate a broad range of quantitative evaluations by leveraging

ndash Crowdsourcing ndash Benchmark data and task taxonomies ndash Automated result interpretation

2 Moving user evaluation from the end to within the software design and implementation process as a means to choose between competitive visual designs

Christian Keimel

copy IRT

Where

What

bull Hybrid broadcast broadband technologies

bull Big Data in broadcasting environment and content-creation

bull Audio-visual QoE assessment (with crowdsourcing)

About me

Christian Keimel Slide 60

Lecturer for Digital Broadcast Engineering

Research Engineer for Media Services and Platforms

copy IRT

Crowdsourcing in large scale optimisation

The human component of Big Data

Moving crowdsourcing to lean-back devices

Crowdsourcing from the couch

Crowdsourced experiments emulating the laboratory

Is it time for new methodologies ignoring the lab

Questionshellip

Christian Keimel Slide 61

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 47: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Matthias Hirth

Matthias Hirth 48 48

About Myself

Matthias Hirth

Research Assistant at University of Wuumlrzburg

httpmatthias-hirthcom

matthiashirthinformatikuni-wuerzburgde

Crowdsourcing

Platform analysis

Platform optimization

Human factors

Use cases

Online Social Networks

Collaboration networks

Structural differences of OSNs

Application of OSN relationships

Interactive applicationsGaming

Challenges of real time interactions

Serious gamingGamification

E-Sports broadcastinglive streaming

Quality of Experience

Monitoring techniques

Quantification techniques

(Crowd-bases) test design

Social interactions

Test methodology Incentive design

Use case

Measurement technique

Identification of

qualification

Matthias Hirth 49 49

Improving Crowdsourcing Technology and the Way We Use it

Crowdsourcing technology

How to transfer well known lab procedures into an uncontrolled

crowdsourcing environment

Which crowdsourcing platform features are required to conduct

successful experiments but are not present yet

Which technical developments (can) foster the usage of

crowdsourcing instead of lab experiments

How can technological developments be used for novel incentive

mechanisms

Usage of Crowdsourcing

How can we conduct reproducible experiments in diverse

crowdsourcing environment

How can we raise the awareness of crowdsourcing employers for the

individuals behind the crowd

Tobias Hoszligfeld

The crowd community in the lab

Tobias Hoszligfeld wwwmaswiwiuni-duede

bull Background Tobias Hoszligfeld

ndash 2004-2014 Research group ldquoFuture Internet Applicationsrdquo Chair of Communication Networks Univ of Wuumlrzburg

ndash Since Sep 2014 Chair of Modeling of Adaptive Systems (head) University of Duisburg-Essen

bull Interests Application of crowdsourcing

ndash User studies on Quality of Experience (QoE) for Internet apps eg video QoE in the crowd amp in the lab methodology reliability statisticshellip [2]

ndash Mobile crowdsourcing network and context measurements eg Wifi offloading incentives privacy recommendation mechanisms etc [1]

bull Interests The crowd community

ndash Crowd = Human beings Social network wisdom of the crowd community

ndash UnderstandingUtilizing the crowd community for successful human computation For which jobs can social interaction improve the quality of work Classification

ndash How can we deploy such a system where interaction is triggered

ndash System design Task design interaction mechanisms incentives

22 Nov 2015

[1] httpdropsdagstuhldeopusvolltexte20134354

[2] httpswww3informatikuni-wuerzburgdeqoewikiqualinetcrowd

Challenges Example of Crowdsourced QoE

Tobias Hoszligfeld wwwmaswiwiuni-duede

task design

incentives

Internet

crowd

community

anonymous

remote

subjects

Internet delivery of

test data and user

response

QoE study bull testing methodology bull QoE influence factors

under investigation bull additional QoE factors bull proper subset of items bull hellip

crowdsourcing setting bull short task duration bull instructions amp training bull feedback amp questions bull hidden influence factors bull reliability of users bull hellip

heterogeneous devices

and Internet access

implementation extrinsic vs intrinsic

motivation

Processing of data

Technical aspects bull server implementation bull client application amp

compatibility bull desired test conditions bull context monitoring bull hellip

test items

lab

scientific methodologies

22 Nov 2015

Jason Jacques

Jason Jacques supervised by Per Ola Kristensson

T

Crowdsourcing Code Perception

jtj21camacuk peopledscamacukjtj21

jtjacquesgmailcom jsonjcouk

PhD Working Title

science

Crowdsourcing Interests W

hat

W

ho

Ho

w

Wh

y jtj21camacuk jsonjcouk science

Radu Jianu

Radu Jianu Assistant Professor

Florida International University (PhD Brown University lsquo12)

Interests

Applications of eye-tracking in visualization tracking what people look at rather than where they look

bull Data intensive large scale eye-tracking experiments bull Gaze-contingent visualizations bull Gaze-driven data recommendation and big data curation

Automating user studies in data visualizations (on next page)

Automating user studies

1 Software designs that can (semi)automate a broad range of quantitative evaluations by leveraging

ndash Crowdsourcing ndash Benchmark data and task taxonomies ndash Automated result interpretation

2 Moving user evaluation from the end to within the software design and implementation process as a means to choose between competitive visual designs

Christian Keimel

copy IRT

Where

What

bull Hybrid broadcast broadband technologies

bull Big Data in broadcasting environment and content-creation

bull Audio-visual QoE assessment (with crowdsourcing)

About me

Christian Keimel Slide 60

Lecturer for Digital Broadcast Engineering

Research Engineer for Media Services and Platforms

copy IRT

Crowdsourcing in large scale optimisation

The human component of Big Data

Moving crowdsourcing to lean-back devices

Crowdsourcing from the couch

Crowdsourced experiments emulating the laboratory

Is it time for new methodologies ignoring the lab

Questionshellip

Christian Keimel Slide 61

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 48: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Matthias Hirth 48 48

About Myself

Matthias Hirth

Research Assistant at University of Wuumlrzburg

httpmatthias-hirthcom

matthiashirthinformatikuni-wuerzburgde

Crowdsourcing

Platform analysis

Platform optimization

Human factors

Use cases

Online Social Networks

Collaboration networks

Structural differences of OSNs

Application of OSN relationships

Interactive applicationsGaming

Challenges of real time interactions

Serious gamingGamification

E-Sports broadcastinglive streaming

Quality of Experience

Monitoring techniques

Quantification techniques

(Crowd-bases) test design

Social interactions

Test methodology Incentive design

Use case

Measurement technique

Identification of

qualification

Matthias Hirth 49 49

Improving Crowdsourcing Technology and the Way We Use it

Crowdsourcing technology

How to transfer well known lab procedures into an uncontrolled

crowdsourcing environment

Which crowdsourcing platform features are required to conduct

successful experiments but are not present yet

Which technical developments (can) foster the usage of

crowdsourcing instead of lab experiments

How can technological developments be used for novel incentive

mechanisms

Usage of Crowdsourcing

How can we conduct reproducible experiments in diverse

crowdsourcing environment

How can we raise the awareness of crowdsourcing employers for the

individuals behind the crowd

Tobias Hoszligfeld

The crowd community in the lab

Tobias Hoszligfeld wwwmaswiwiuni-duede

bull Background Tobias Hoszligfeld

ndash 2004-2014 Research group ldquoFuture Internet Applicationsrdquo Chair of Communication Networks Univ of Wuumlrzburg

ndash Since Sep 2014 Chair of Modeling of Adaptive Systems (head) University of Duisburg-Essen

bull Interests Application of crowdsourcing

ndash User studies on Quality of Experience (QoE) for Internet apps eg video QoE in the crowd amp in the lab methodology reliability statisticshellip [2]

ndash Mobile crowdsourcing network and context measurements eg Wifi offloading incentives privacy recommendation mechanisms etc [1]

bull Interests The crowd community

ndash Crowd = Human beings Social network wisdom of the crowd community

ndash UnderstandingUtilizing the crowd community for successful human computation For which jobs can social interaction improve the quality of work Classification

ndash How can we deploy such a system where interaction is triggered

ndash System design Task design interaction mechanisms incentives

22 Nov 2015

[1] httpdropsdagstuhldeopusvolltexte20134354

[2] httpswww3informatikuni-wuerzburgdeqoewikiqualinetcrowd

Challenges Example of Crowdsourced QoE

Tobias Hoszligfeld wwwmaswiwiuni-duede

task design

incentives

Internet

crowd

community

anonymous

remote

subjects

Internet delivery of

test data and user

response

QoE study bull testing methodology bull QoE influence factors

under investigation bull additional QoE factors bull proper subset of items bull hellip

crowdsourcing setting bull short task duration bull instructions amp training bull feedback amp questions bull hidden influence factors bull reliability of users bull hellip

heterogeneous devices

and Internet access

implementation extrinsic vs intrinsic

motivation

Processing of data

Technical aspects bull server implementation bull client application amp

compatibility bull desired test conditions bull context monitoring bull hellip

test items

lab

scientific methodologies

22 Nov 2015

Jason Jacques

Jason Jacques supervised by Per Ola Kristensson

T

Crowdsourcing Code Perception

jtj21camacuk peopledscamacukjtj21

jtjacquesgmailcom jsonjcouk

PhD Working Title

science

Crowdsourcing Interests W

hat

W

ho

Ho

w

Wh

y jtj21camacuk jsonjcouk science

Radu Jianu

Radu Jianu Assistant Professor

Florida International University (PhD Brown University lsquo12)

Interests

Applications of eye-tracking in visualization tracking what people look at rather than where they look

bull Data intensive large scale eye-tracking experiments bull Gaze-contingent visualizations bull Gaze-driven data recommendation and big data curation

Automating user studies in data visualizations (on next page)

Automating user studies

1 Software designs that can (semi)automate a broad range of quantitative evaluations by leveraging

ndash Crowdsourcing ndash Benchmark data and task taxonomies ndash Automated result interpretation

2 Moving user evaluation from the end to within the software design and implementation process as a means to choose between competitive visual designs

Christian Keimel

copy IRT

Where

What

bull Hybrid broadcast broadband technologies

bull Big Data in broadcasting environment and content-creation

bull Audio-visual QoE assessment (with crowdsourcing)

About me

Christian Keimel Slide 60

Lecturer for Digital Broadcast Engineering

Research Engineer for Media Services and Platforms

copy IRT

Crowdsourcing in large scale optimisation

The human component of Big Data

Moving crowdsourcing to lean-back devices

Crowdsourcing from the couch

Crowdsourced experiments emulating the laboratory

Is it time for new methodologies ignoring the lab

Questionshellip

Christian Keimel Slide 61

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 49: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Matthias Hirth 49 49

Improving Crowdsourcing Technology and the Way We Use it

Crowdsourcing technology

How to transfer well known lab procedures into an uncontrolled

crowdsourcing environment

Which crowdsourcing platform features are required to conduct

successful experiments but are not present yet

Which technical developments (can) foster the usage of

crowdsourcing instead of lab experiments

How can technological developments be used for novel incentive

mechanisms

Usage of Crowdsourcing

How can we conduct reproducible experiments in diverse

crowdsourcing environment

How can we raise the awareness of crowdsourcing employers for the

individuals behind the crowd

Tobias Hoszligfeld

The crowd community in the lab

Tobias Hoszligfeld wwwmaswiwiuni-duede

bull Background Tobias Hoszligfeld

ndash 2004-2014 Research group ldquoFuture Internet Applicationsrdquo Chair of Communication Networks Univ of Wuumlrzburg

ndash Since Sep 2014 Chair of Modeling of Adaptive Systems (head) University of Duisburg-Essen

bull Interests Application of crowdsourcing

ndash User studies on Quality of Experience (QoE) for Internet apps eg video QoE in the crowd amp in the lab methodology reliability statisticshellip [2]

ndash Mobile crowdsourcing network and context measurements eg Wifi offloading incentives privacy recommendation mechanisms etc [1]

bull Interests The crowd community

ndash Crowd = Human beings Social network wisdom of the crowd community

ndash UnderstandingUtilizing the crowd community for successful human computation For which jobs can social interaction improve the quality of work Classification

ndash How can we deploy such a system where interaction is triggered

ndash System design Task design interaction mechanisms incentives

22 Nov 2015

[1] httpdropsdagstuhldeopusvolltexte20134354

[2] httpswww3informatikuni-wuerzburgdeqoewikiqualinetcrowd

Challenges Example of Crowdsourced QoE

Tobias Hoszligfeld wwwmaswiwiuni-duede

task design

incentives

Internet

crowd

community

anonymous

remote

subjects

Internet delivery of

test data and user

response

QoE study bull testing methodology bull QoE influence factors

under investigation bull additional QoE factors bull proper subset of items bull hellip

crowdsourcing setting bull short task duration bull instructions amp training bull feedback amp questions bull hidden influence factors bull reliability of users bull hellip

heterogeneous devices

and Internet access

implementation extrinsic vs intrinsic

motivation

Processing of data

Technical aspects bull server implementation bull client application amp

compatibility bull desired test conditions bull context monitoring bull hellip

test items

lab

scientific methodologies

22 Nov 2015

Jason Jacques

Jason Jacques supervised by Per Ola Kristensson

T

Crowdsourcing Code Perception

jtj21camacuk peopledscamacukjtj21

jtjacquesgmailcom jsonjcouk

PhD Working Title

science

Crowdsourcing Interests W

hat

W

ho

Ho

w

Wh

y jtj21camacuk jsonjcouk science

Radu Jianu

Radu Jianu Assistant Professor

Florida International University (PhD Brown University lsquo12)

Interests

Applications of eye-tracking in visualization tracking what people look at rather than where they look

bull Data intensive large scale eye-tracking experiments bull Gaze-contingent visualizations bull Gaze-driven data recommendation and big data curation

Automating user studies in data visualizations (on next page)

Automating user studies

1 Software designs that can (semi)automate a broad range of quantitative evaluations by leveraging

ndash Crowdsourcing ndash Benchmark data and task taxonomies ndash Automated result interpretation

2 Moving user evaluation from the end to within the software design and implementation process as a means to choose between competitive visual designs

Christian Keimel

copy IRT

Where

What

bull Hybrid broadcast broadband technologies

bull Big Data in broadcasting environment and content-creation

bull Audio-visual QoE assessment (with crowdsourcing)

About me

Christian Keimel Slide 60

Lecturer for Digital Broadcast Engineering

Research Engineer for Media Services and Platforms

copy IRT

Crowdsourcing in large scale optimisation

The human component of Big Data

Moving crowdsourcing to lean-back devices

Crowdsourcing from the couch

Crowdsourced experiments emulating the laboratory

Is it time for new methodologies ignoring the lab

Questionshellip

Christian Keimel Slide 61

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 50: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Tobias Hoszligfeld

The crowd community in the lab

Tobias Hoszligfeld wwwmaswiwiuni-duede

bull Background Tobias Hoszligfeld

ndash 2004-2014 Research group ldquoFuture Internet Applicationsrdquo Chair of Communication Networks Univ of Wuumlrzburg

ndash Since Sep 2014 Chair of Modeling of Adaptive Systems (head) University of Duisburg-Essen

bull Interests Application of crowdsourcing

ndash User studies on Quality of Experience (QoE) for Internet apps eg video QoE in the crowd amp in the lab methodology reliability statisticshellip [2]

ndash Mobile crowdsourcing network and context measurements eg Wifi offloading incentives privacy recommendation mechanisms etc [1]

bull Interests The crowd community

ndash Crowd = Human beings Social network wisdom of the crowd community

ndash UnderstandingUtilizing the crowd community for successful human computation For which jobs can social interaction improve the quality of work Classification

ndash How can we deploy such a system where interaction is triggered

ndash System design Task design interaction mechanisms incentives

22 Nov 2015

[1] httpdropsdagstuhldeopusvolltexte20134354

[2] httpswww3informatikuni-wuerzburgdeqoewikiqualinetcrowd

Challenges Example of Crowdsourced QoE

Tobias Hoszligfeld wwwmaswiwiuni-duede

task design

incentives

Internet

crowd

community

anonymous

remote

subjects

Internet delivery of

test data and user

response

QoE study bull testing methodology bull QoE influence factors

under investigation bull additional QoE factors bull proper subset of items bull hellip

crowdsourcing setting bull short task duration bull instructions amp training bull feedback amp questions bull hidden influence factors bull reliability of users bull hellip

heterogeneous devices

and Internet access

implementation extrinsic vs intrinsic

motivation

Processing of data

Technical aspects bull server implementation bull client application amp

compatibility bull desired test conditions bull context monitoring bull hellip

test items

lab

scientific methodologies

22 Nov 2015

Jason Jacques

Jason Jacques supervised by Per Ola Kristensson

T

Crowdsourcing Code Perception

jtj21camacuk peopledscamacukjtj21

jtjacquesgmailcom jsonjcouk

PhD Working Title

science

Crowdsourcing Interests W

hat

W

ho

Ho

w

Wh

y jtj21camacuk jsonjcouk science

Radu Jianu

Radu Jianu Assistant Professor

Florida International University (PhD Brown University lsquo12)

Interests

Applications of eye-tracking in visualization tracking what people look at rather than where they look

bull Data intensive large scale eye-tracking experiments bull Gaze-contingent visualizations bull Gaze-driven data recommendation and big data curation

Automating user studies in data visualizations (on next page)

Automating user studies

1 Software designs that can (semi)automate a broad range of quantitative evaluations by leveraging

ndash Crowdsourcing ndash Benchmark data and task taxonomies ndash Automated result interpretation

2 Moving user evaluation from the end to within the software design and implementation process as a means to choose between competitive visual designs

Christian Keimel

copy IRT

Where

What

bull Hybrid broadcast broadband technologies

bull Big Data in broadcasting environment and content-creation

bull Audio-visual QoE assessment (with crowdsourcing)

About me

Christian Keimel Slide 60

Lecturer for Digital Broadcast Engineering

Research Engineer for Media Services and Platforms

copy IRT

Crowdsourcing in large scale optimisation

The human component of Big Data

Moving crowdsourcing to lean-back devices

Crowdsourcing from the couch

Crowdsourced experiments emulating the laboratory

Is it time for new methodologies ignoring the lab

Questionshellip

Christian Keimel Slide 61

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 51: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

The crowd community in the lab

Tobias Hoszligfeld wwwmaswiwiuni-duede

bull Background Tobias Hoszligfeld

ndash 2004-2014 Research group ldquoFuture Internet Applicationsrdquo Chair of Communication Networks Univ of Wuumlrzburg

ndash Since Sep 2014 Chair of Modeling of Adaptive Systems (head) University of Duisburg-Essen

bull Interests Application of crowdsourcing

ndash User studies on Quality of Experience (QoE) for Internet apps eg video QoE in the crowd amp in the lab methodology reliability statisticshellip [2]

ndash Mobile crowdsourcing network and context measurements eg Wifi offloading incentives privacy recommendation mechanisms etc [1]

bull Interests The crowd community

ndash Crowd = Human beings Social network wisdom of the crowd community

ndash UnderstandingUtilizing the crowd community for successful human computation For which jobs can social interaction improve the quality of work Classification

ndash How can we deploy such a system where interaction is triggered

ndash System design Task design interaction mechanisms incentives

22 Nov 2015

[1] httpdropsdagstuhldeopusvolltexte20134354

[2] httpswww3informatikuni-wuerzburgdeqoewikiqualinetcrowd

Challenges Example of Crowdsourced QoE

Tobias Hoszligfeld wwwmaswiwiuni-duede

task design

incentives

Internet

crowd

community

anonymous

remote

subjects

Internet delivery of

test data and user

response

QoE study bull testing methodology bull QoE influence factors

under investigation bull additional QoE factors bull proper subset of items bull hellip

crowdsourcing setting bull short task duration bull instructions amp training bull feedback amp questions bull hidden influence factors bull reliability of users bull hellip

heterogeneous devices

and Internet access

implementation extrinsic vs intrinsic

motivation

Processing of data

Technical aspects bull server implementation bull client application amp

compatibility bull desired test conditions bull context monitoring bull hellip

test items

lab

scientific methodologies

22 Nov 2015

Jason Jacques

Jason Jacques supervised by Per Ola Kristensson

T

Crowdsourcing Code Perception

jtj21camacuk peopledscamacukjtj21

jtjacquesgmailcom jsonjcouk

PhD Working Title

science

Crowdsourcing Interests W

hat

W

ho

Ho

w

Wh

y jtj21camacuk jsonjcouk science

Radu Jianu

Radu Jianu Assistant Professor

Florida International University (PhD Brown University lsquo12)

Interests

Applications of eye-tracking in visualization tracking what people look at rather than where they look

bull Data intensive large scale eye-tracking experiments bull Gaze-contingent visualizations bull Gaze-driven data recommendation and big data curation

Automating user studies in data visualizations (on next page)

Automating user studies

1 Software designs that can (semi)automate a broad range of quantitative evaluations by leveraging

ndash Crowdsourcing ndash Benchmark data and task taxonomies ndash Automated result interpretation

2 Moving user evaluation from the end to within the software design and implementation process as a means to choose between competitive visual designs

Christian Keimel

copy IRT

Where

What

bull Hybrid broadcast broadband technologies

bull Big Data in broadcasting environment and content-creation

bull Audio-visual QoE assessment (with crowdsourcing)

About me

Christian Keimel Slide 60

Lecturer for Digital Broadcast Engineering

Research Engineer for Media Services and Platforms

copy IRT

Crowdsourcing in large scale optimisation

The human component of Big Data

Moving crowdsourcing to lean-back devices

Crowdsourcing from the couch

Crowdsourced experiments emulating the laboratory

Is it time for new methodologies ignoring the lab

Questionshellip

Christian Keimel Slide 61

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 52: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Challenges Example of Crowdsourced QoE

Tobias Hoszligfeld wwwmaswiwiuni-duede

task design

incentives

Internet

crowd

community

anonymous

remote

subjects

Internet delivery of

test data and user

response

QoE study bull testing methodology bull QoE influence factors

under investigation bull additional QoE factors bull proper subset of items bull hellip

crowdsourcing setting bull short task duration bull instructions amp training bull feedback amp questions bull hidden influence factors bull reliability of users bull hellip

heterogeneous devices

and Internet access

implementation extrinsic vs intrinsic

motivation

Processing of data

Technical aspects bull server implementation bull client application amp

compatibility bull desired test conditions bull context monitoring bull hellip

test items

lab

scientific methodologies

22 Nov 2015

Jason Jacques

Jason Jacques supervised by Per Ola Kristensson

T

Crowdsourcing Code Perception

jtj21camacuk peopledscamacukjtj21

jtjacquesgmailcom jsonjcouk

PhD Working Title

science

Crowdsourcing Interests W

hat

W

ho

Ho

w

Wh

y jtj21camacuk jsonjcouk science

Radu Jianu

Radu Jianu Assistant Professor

Florida International University (PhD Brown University lsquo12)

Interests

Applications of eye-tracking in visualization tracking what people look at rather than where they look

bull Data intensive large scale eye-tracking experiments bull Gaze-contingent visualizations bull Gaze-driven data recommendation and big data curation

Automating user studies in data visualizations (on next page)

Automating user studies

1 Software designs that can (semi)automate a broad range of quantitative evaluations by leveraging

ndash Crowdsourcing ndash Benchmark data and task taxonomies ndash Automated result interpretation

2 Moving user evaluation from the end to within the software design and implementation process as a means to choose between competitive visual designs

Christian Keimel

copy IRT

Where

What

bull Hybrid broadcast broadband technologies

bull Big Data in broadcasting environment and content-creation

bull Audio-visual QoE assessment (with crowdsourcing)

About me

Christian Keimel Slide 60

Lecturer for Digital Broadcast Engineering

Research Engineer for Media Services and Platforms

copy IRT

Crowdsourcing in large scale optimisation

The human component of Big Data

Moving crowdsourcing to lean-back devices

Crowdsourcing from the couch

Crowdsourced experiments emulating the laboratory

Is it time for new methodologies ignoring the lab

Questionshellip

Christian Keimel Slide 61

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 53: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Jason Jacques

Jason Jacques supervised by Per Ola Kristensson

T

Crowdsourcing Code Perception

jtj21camacuk peopledscamacukjtj21

jtjacquesgmailcom jsonjcouk

PhD Working Title

science

Crowdsourcing Interests W

hat

W

ho

Ho

w

Wh

y jtj21camacuk jsonjcouk science

Radu Jianu

Radu Jianu Assistant Professor

Florida International University (PhD Brown University lsquo12)

Interests

Applications of eye-tracking in visualization tracking what people look at rather than where they look

bull Data intensive large scale eye-tracking experiments bull Gaze-contingent visualizations bull Gaze-driven data recommendation and big data curation

Automating user studies in data visualizations (on next page)

Automating user studies

1 Software designs that can (semi)automate a broad range of quantitative evaluations by leveraging

ndash Crowdsourcing ndash Benchmark data and task taxonomies ndash Automated result interpretation

2 Moving user evaluation from the end to within the software design and implementation process as a means to choose between competitive visual designs

Christian Keimel

copy IRT

Where

What

bull Hybrid broadcast broadband technologies

bull Big Data in broadcasting environment and content-creation

bull Audio-visual QoE assessment (with crowdsourcing)

About me

Christian Keimel Slide 60

Lecturer for Digital Broadcast Engineering

Research Engineer for Media Services and Platforms

copy IRT

Crowdsourcing in large scale optimisation

The human component of Big Data

Moving crowdsourcing to lean-back devices

Crowdsourcing from the couch

Crowdsourced experiments emulating the laboratory

Is it time for new methodologies ignoring the lab

Questionshellip

Christian Keimel Slide 61

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 54: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Jason Jacques supervised by Per Ola Kristensson

T

Crowdsourcing Code Perception

jtj21camacuk peopledscamacukjtj21

jtjacquesgmailcom jsonjcouk

PhD Working Title

science

Crowdsourcing Interests W

hat

W

ho

Ho

w

Wh

y jtj21camacuk jsonjcouk science

Radu Jianu

Radu Jianu Assistant Professor

Florida International University (PhD Brown University lsquo12)

Interests

Applications of eye-tracking in visualization tracking what people look at rather than where they look

bull Data intensive large scale eye-tracking experiments bull Gaze-contingent visualizations bull Gaze-driven data recommendation and big data curation

Automating user studies in data visualizations (on next page)

Automating user studies

1 Software designs that can (semi)automate a broad range of quantitative evaluations by leveraging

ndash Crowdsourcing ndash Benchmark data and task taxonomies ndash Automated result interpretation

2 Moving user evaluation from the end to within the software design and implementation process as a means to choose between competitive visual designs

Christian Keimel

copy IRT

Where

What

bull Hybrid broadcast broadband technologies

bull Big Data in broadcasting environment and content-creation

bull Audio-visual QoE assessment (with crowdsourcing)

About me

Christian Keimel Slide 60

Lecturer for Digital Broadcast Engineering

Research Engineer for Media Services and Platforms

copy IRT

Crowdsourcing in large scale optimisation

The human component of Big Data

Moving crowdsourcing to lean-back devices

Crowdsourcing from the couch

Crowdsourced experiments emulating the laboratory

Is it time for new methodologies ignoring the lab

Questionshellip

Christian Keimel Slide 61

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 55: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Crowdsourcing Interests W

hat

W

ho

Ho

w

Wh

y jtj21camacuk jsonjcouk science

Radu Jianu

Radu Jianu Assistant Professor

Florida International University (PhD Brown University lsquo12)

Interests

Applications of eye-tracking in visualization tracking what people look at rather than where they look

bull Data intensive large scale eye-tracking experiments bull Gaze-contingent visualizations bull Gaze-driven data recommendation and big data curation

Automating user studies in data visualizations (on next page)

Automating user studies

1 Software designs that can (semi)automate a broad range of quantitative evaluations by leveraging

ndash Crowdsourcing ndash Benchmark data and task taxonomies ndash Automated result interpretation

2 Moving user evaluation from the end to within the software design and implementation process as a means to choose between competitive visual designs

Christian Keimel

copy IRT

Where

What

bull Hybrid broadcast broadband technologies

bull Big Data in broadcasting environment and content-creation

bull Audio-visual QoE assessment (with crowdsourcing)

About me

Christian Keimel Slide 60

Lecturer for Digital Broadcast Engineering

Research Engineer for Media Services and Platforms

copy IRT

Crowdsourcing in large scale optimisation

The human component of Big Data

Moving crowdsourcing to lean-back devices

Crowdsourcing from the couch

Crowdsourced experiments emulating the laboratory

Is it time for new methodologies ignoring the lab

Questionshellip

Christian Keimel Slide 61

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 56: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Radu Jianu

Radu Jianu Assistant Professor

Florida International University (PhD Brown University lsquo12)

Interests

Applications of eye-tracking in visualization tracking what people look at rather than where they look

bull Data intensive large scale eye-tracking experiments bull Gaze-contingent visualizations bull Gaze-driven data recommendation and big data curation

Automating user studies in data visualizations (on next page)

Automating user studies

1 Software designs that can (semi)automate a broad range of quantitative evaluations by leveraging

ndash Crowdsourcing ndash Benchmark data and task taxonomies ndash Automated result interpretation

2 Moving user evaluation from the end to within the software design and implementation process as a means to choose between competitive visual designs

Christian Keimel

copy IRT

Where

What

bull Hybrid broadcast broadband technologies

bull Big Data in broadcasting environment and content-creation

bull Audio-visual QoE assessment (with crowdsourcing)

About me

Christian Keimel Slide 60

Lecturer for Digital Broadcast Engineering

Research Engineer for Media Services and Platforms

copy IRT

Crowdsourcing in large scale optimisation

The human component of Big Data

Moving crowdsourcing to lean-back devices

Crowdsourcing from the couch

Crowdsourced experiments emulating the laboratory

Is it time for new methodologies ignoring the lab

Questionshellip

Christian Keimel Slide 61

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 57: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Radu Jianu Assistant Professor

Florida International University (PhD Brown University lsquo12)

Interests

Applications of eye-tracking in visualization tracking what people look at rather than where they look

bull Data intensive large scale eye-tracking experiments bull Gaze-contingent visualizations bull Gaze-driven data recommendation and big data curation

Automating user studies in data visualizations (on next page)

Automating user studies

1 Software designs that can (semi)automate a broad range of quantitative evaluations by leveraging

ndash Crowdsourcing ndash Benchmark data and task taxonomies ndash Automated result interpretation

2 Moving user evaluation from the end to within the software design and implementation process as a means to choose between competitive visual designs

Christian Keimel

copy IRT

Where

What

bull Hybrid broadcast broadband technologies

bull Big Data in broadcasting environment and content-creation

bull Audio-visual QoE assessment (with crowdsourcing)

About me

Christian Keimel Slide 60

Lecturer for Digital Broadcast Engineering

Research Engineer for Media Services and Platforms

copy IRT

Crowdsourcing in large scale optimisation

The human component of Big Data

Moving crowdsourcing to lean-back devices

Crowdsourcing from the couch

Crowdsourced experiments emulating the laboratory

Is it time for new methodologies ignoring the lab

Questionshellip

Christian Keimel Slide 61

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 58: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Automating user studies

1 Software designs that can (semi)automate a broad range of quantitative evaluations by leveraging

ndash Crowdsourcing ndash Benchmark data and task taxonomies ndash Automated result interpretation

2 Moving user evaluation from the end to within the software design and implementation process as a means to choose between competitive visual designs

Christian Keimel

copy IRT

Where

What

bull Hybrid broadcast broadband technologies

bull Big Data in broadcasting environment and content-creation

bull Audio-visual QoE assessment (with crowdsourcing)

About me

Christian Keimel Slide 60

Lecturer for Digital Broadcast Engineering

Research Engineer for Media Services and Platforms

copy IRT

Crowdsourcing in large scale optimisation

The human component of Big Data

Moving crowdsourcing to lean-back devices

Crowdsourcing from the couch

Crowdsourced experiments emulating the laboratory

Is it time for new methodologies ignoring the lab

Questionshellip

Christian Keimel Slide 61

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 59: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Christian Keimel

copy IRT

Where

What

bull Hybrid broadcast broadband technologies

bull Big Data in broadcasting environment and content-creation

bull Audio-visual QoE assessment (with crowdsourcing)

About me

Christian Keimel Slide 60

Lecturer for Digital Broadcast Engineering

Research Engineer for Media Services and Platforms

copy IRT

Crowdsourcing in large scale optimisation

The human component of Big Data

Moving crowdsourcing to lean-back devices

Crowdsourcing from the couch

Crowdsourced experiments emulating the laboratory

Is it time for new methodologies ignoring the lab

Questionshellip

Christian Keimel Slide 61

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 60: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

copy IRT

Where

What

bull Hybrid broadcast broadband technologies

bull Big Data in broadcasting environment and content-creation

bull Audio-visual QoE assessment (with crowdsourcing)

About me

Christian Keimel Slide 60

Lecturer for Digital Broadcast Engineering

Research Engineer for Media Services and Platforms

copy IRT

Crowdsourcing in large scale optimisation

The human component of Big Data

Moving crowdsourcing to lean-back devices

Crowdsourcing from the couch

Crowdsourced experiments emulating the laboratory

Is it time for new methodologies ignoring the lab

Questionshellip

Christian Keimel Slide 61

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 61: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

copy IRT

Crowdsourcing in large scale optimisation

The human component of Big Data

Moving crowdsourcing to lean-back devices

Crowdsourcing from the couch

Crowdsourced experiments emulating the laboratory

Is it time for new methodologies ignoring the lab

Questionshellip

Christian Keimel Slide 61

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 62: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Andreas Kerren

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 63: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Seminar 15481

Andreas Kerren kerrenacmorg

bull Affiliation

ndash Professor in CS at Linnaeus University Sweden

ndash Head of the ISOVIS Research Group

ndash httpcslnuseisovis

bull Current research interests

ndash Multivariate network analysis navigating in networks

ndash Visual text analytics

ndash BCI-supported evaluation

ndash Collaborativeadaptive information visualization

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 64: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Seminar 15481

Andreas Kerren kerrenacmorg

bull Irsquom interested in to learn more on or to discuss

ndash Which evaluation tasks for interactive visualizations are

appropriate for crowdsourcing

ndash More concrete is evaluation in the crowd suitable for

testing more complex interactions

bull It seems that this is problematic due to the motivation of

the people

ndash Which platform is the best for my purposes and what are

the pros and cons of each

ndash Practical guidelines to launch such experiments

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 65: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Stephen Kobourov

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 66: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Background

Stephen Kobourov

Computer Science

University of Arizona

Research Interests

information visualization

graph drawing

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 67: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Topics

Point-clouds graphs and maps

Edge crossings in graph drawing

Straight-line and curved edges

Semantic wordclouds

Cartograms

In progress

- memorability

- engagement

- enjoyment

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 68: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Bongshin Lee

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 69: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Bongshin Lee

Senior Researcher Microsoft Research

PhD University of Maryland College Park 2006

Analyze Share Collect

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 70: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Understand and share the experiences in running crowdsourcing experiments

Discuss the strengths and limitations of crowdsourcing experiments

Formulate the future research directions in crowdsourcing experiments

Interests

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 71: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

David Martin

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 72: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Crowdsourcing and Human-Centred Experiments

bull David Martin

ndash Senior scientist Xerox Research Centre Europe

ndash Sociologist specializing in ethnographic (detailed qualitative observational) studies of social organisation

bull Studies for design and innovation

bull Studies of design and innovation

bull Looking at use and impacts of technology in society

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 73: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Crowdsourcing Interests

bull Understanding the use understanding and impact of crowdsourcing on crowdsourcing workers

bull Crowdsourcing as a new form of work ndash ie as a global form of technologically enabled outsourcing and as precarious work

bull Designing technologies to aid and empower crowdsourcing workers

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 74: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Andrea Mauri

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 75: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 75

bull Third year PhD Student at Politecnico di Milano

bull Thesis ldquoMethodologies for the development of crowd and social-based

applicationsrdquo

Defined a design methodology a specification paradigm deploying and monitoring

crowd-based applications

bull Interests

Crowdsourcing and human computation

Social media content analysis

Model driven engineering

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 76: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Dipartimento di Elettronica Informazione e Bioingegneria

Andrea Mauri 76

Crowdsourcing Experiments in Human-Computer Interaction Visualization and

Applied PerceptionGraphics

bull How the crowd can be involved in the development of a modeling language

Crowdsourcing Platforms vs The Laboratory

Scientifically Rigorous Methodologies

bull Best practices

Ethics in Experiments

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 77: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Fintan McGee

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 78: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

FINTAN MCGEE

Graph amp Information Visualization

bull Lab Based Evaluation

bull Edge Bundling

bull Matrix VS Node link for community visualization

bull Biological Data Visualization

Phd from Trinity College Dublin (2013)

bull Graphics Vision and Visualization Group (GV2)

eScience Unit - Luxembourg Institute of Science and

Technology

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 79: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

SEMINAR INTERESTS

Obtaining suitable populations

bull For application domain experts

bull For general visualization approaches

Crowdsourcing best practices

Crowdsourcing evaluation of visualization within

mobile applications

Insights

bull Not available in lab experiments

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 80: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Luana Micallef

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 81: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Luana Micallef Postdoctoral Researcher

Research Interests

Information Visualization Biological Data Visualization Set Visualization

Diagram Layout Optimization Visual Design and Analytics Human-Computer Interaction

Ubiquitous

Interaction

Group

Probabilistic

Machine

Learning

Group

Giulio Jacucci Samuel Kaski Aalto University and University of Helsinki

Honorary Research Fellow [2014ndash2017]

Research Fellow [2013ndash2014]

PhD [2009ndash2013] Visualizing Set Relations and Cardinalities Using Venn and

Euler Diagrams Advisor Peter Rodgers

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 82: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Luana Micallef Interests Related to Crowdsourcing and Human-Centred Experiments

Using Amazon Mechanical Turk (MTurk)

evaluated the effectiveness of different visualizations

eg set visualization techniques for different tasks

Next consider crowdsourcing for the evaluation of

interactive visualizations user interfaces and information retrieval systems

learn more about

other crowdsourcing technologies besides MTurk

characteristics of the crowd and ethics of crowdsourcing in research

Also adopted different spammer-catching techniques and

created templates for crowdsourced web experiments

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 83: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Babak Naderi

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 84: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Babak Naderi Research Scientist PhD Candidate

Quality and Usability Lab

Telekom Innovation Laboratories

Technische Universitaumlt Berlin

Background Study Bachelorrsquos degree Software Engineering Master lsquos

Geodesy and Geoinformation Science

Activities PhD topic Motivation of Crowd workers on Online Crowdsourcing

Micro-Task Platforms

Focus on Measuring motivation of crowd workers amp evaluating factors influencing it Influence of motivation on performance Design guideline for lsquoproperrsquo crowdsourcing task Difference between Lab and crowdsourcing studies

Crowdee Mobile Crowdsourcing platform provided by QUL TU-Berlin

Turkmotion Initiative that aims to support the crowdworkersrsquo community Share your vote - enjoy your work

Interests Motivation What can change initial Motivation (External -gt

Intrinsic) Look at the motivation in details (not a black box) how does it change

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 85: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Martin Noumlllenburg

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 86: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Helen Purchase

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 87: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Empirical studies in information visualisation and aesthetic design

Sketched graph layout

Graph drawing aesthetics

static graphs dynamic graphs curved edges cascades

Aesthetic interface design and complexity

Helen Purchase

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 88: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Interesting questions

It is really lsquobetterrsquo to get crowd-sourced empirical data If so why and under what conditions

Do our empirical design methods need to change when we use crowd-sourced empirical data

If so why and how

Do our analysis methods need to change when we use crowd-sourced empirical data

If so why and how

Who are our participants Why do they do it Does it matter If so why and how does this affect what we do

Helen Purchase

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 89: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Judith Redi

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 90: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Assistant Professor

Multimedia

Computing Group

Delft

Genoa

Ivrea

Eindhoven

Sophia

Antipolis

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 91: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

QOE CONTROL

Configuration of

technology

settings

Quality

restoration

User-centered

technology

design

Low

quality

Q = 03

Quality assessment Quality Preservation

Model of

user preferences

Subjective Testing

93

Usually lab based time consuming

limited ecological

validity

Can crowdsourcing help

Can we obtain similar results in a CS and a contolled

Lab setting for QoE measurements How

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 92: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Peter Rodgers

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 93: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Peter Rodgers

Reader amp Director of Research

School of Computing

University of Kent

UK

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 94: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Research Interests Developing new Set amp Network Visualizations

Empirical Evaluation of Information Visualizations

Currently using MTurk amp Laboratory based techniques

httpwwwcskentacukpeoplestaffpjrlinear With Gem Stapleton and Peter Chapman

httpwwwcskentacukpeoplestaffpjrEulerVennCirclesEulerVennApplethtml With Stirling Chow

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 95: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Dietmar Saupe

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 96: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Universitaumlt Konstanz

Chair for Multimedia Signal Processing University of Konstanz Germany Prof Dr Dietmar Saupe

Current Research Projects

bull Powerbike Data acquisition analysis modeling optimization of

performance in endurance sports (competitive cycling)

bull BrainCycles Recordinganalysing brain activity of Parkinson

patients during walkingcycling

bull IVQA Objectivesubjective assessment of imagevideo quality

using crowd sourcing eyetracking

bull Automated detection of archaeological sites in high resolution

remotely sensed imagery

Dagstuhl Seminar Crowdsourcing 23-27-112015 98 22112015

CV

Education in (applied) mathematics

Diploma (1979)

Dr rer nat (1982)

Habilitation (1993)

Past research

Numerical continuation methods

Chaos and fractals

Fractal image compression

Computer graphics

Visualization

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 97: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Universitaumlt Konstanz

Crowd Sourcing for Image and Video Quality Assessment

Dagstuhl Seminar Crowdsourcing 23-27-112015 99 22112015

Consumer video quality

Quality indices

- Overall visual

- Motion

- Color

- Comfort

Limitations of current subjectiveobjective assessment

bull Video databases have narrow spectrum of content

bull only with controlled artificial distortions (compression blurring )

Can be overcome by crowd sourcing

bull Web harvesting for thousands of videos (diversity)

bull with bdquonaturalldquo degradations (authenticity)

Challenges

bull Only a single version per film

bull No ground truth available (limits to no-reference VQA)

bull The usual limitations of crowd sourcing

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 98: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Ognjen Šćekić

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 99: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Ognjen Šćekić Research assistant Distributed Systems Group TU Wien Austria httpdsgtuwienacat

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

Research Interests bull Rewarding and incentives in

Social Computing bull Human-Provided Services (HPS) bull Collective Adaptive Systems (CAS) bull Service-Oriented Computing

Currently involved in research of hybrid human-machine socio-technical systems (EU FP7 SmartSociety project)

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 100: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Dagstuhl Seminar 15481 November 22ndash27 2015

Evaluation in the Crowd Crowdsourcing and Human-Centered Experiments

bull SmartSociety ndash platform for crowdsourcing complex collaborative tasks to hybrid collectives

bull Expecting to use the platform for running experiments and studies involving human participants negotiations agreement controllability incentives

bull Interested in experiences of other participants and discussing ethical questions

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 101: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Paolo Simonetto

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 102: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Paolo Simonetto

Studied worked at

bull University of Padua

bull BRC University of Glasgow

bull LaBRI University of Bordeaux

bull GAMA University of Arizona

Topics

bull Visualization of clustered graphs

bull Generation of Euler diagrams

bull Graph drawing with containment constraints

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 103: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

How not to misuse crowdsourcing

Crowdsourcing vs Standard

bull Design time equal +

bull Supervision time none +

bull Physical presence nah +

bull Analysis same = _________

Letrsquos make it clear

bull When not to use it

bull How to use it properly

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 104: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Tatiana von Landesberger

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 105: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Tatiana von Landesberger

Head of Junior Research Group

TU Darmstadt Germany

Research interests Visual Analysis of Networks

Visual Analysis of Spatio-temporal data

Visual Analysis of Medical Data

Applications

Finance

Biology

Transportation

Social Media (News Twitter)

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 106: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Interests

Visualization for the masses

Crowdsourcing data

for and in visualization

Evaluation of visualization

for and by the masses

httptagesnetzwerkde

wwwviphyorg

MobilityGraphs

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 107: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Ina Wechsung

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 108: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Ina Wechsung

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 109: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Michael Wybrow

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 110: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Michael Wybrow

bull Lecturer ndash Faculty of IT Monash University

bull Research bull Geometric layout constraints bull Diagram amp document layout bull Connector routing bull Information visualisation bull Interaction design and HCI

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 111: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Michael Wybrow

bull Some relevant interests to this seminar

bull PCs and tablets are inherently distracting How focused are crowdsourced participants

bull How can we determine a participantrsquos intentions from their interactions with an online system

bull The value of recording interactions for later replay

bull ResearchKit Applersquos framework for building mobile apps for collecting crowdsourced study data

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 112: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Michelle Zhou

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 113: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

1999 2008 2010 2009

T J Watson

China Research

2014

Almaden Watson Grp

Columbia

Smart Visualization for Conversational Agents

Visual Analytics Concierge

Psychology-based People Analytics

Cognitive Personal Agent Michelle Zhou

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality

Page 114: 1. Daniel Archambaultmaterials.dagstuhl.de/files/15/15481/15481.SWM1.Other.pdf · Platform optimization Human factors Use cases Online Social Networks Collaboration netwo rks Structural

Getting to Know the Crowd

bull What are they like

ndash What motivates them

ndash How trustworthy are they

bull How to get to know them better

ndash Psychometrics tests

ndash Automated analysis

Motivations

Decision-Making

Style

Personality