extracting, aggregating and visualizing events from text

Post on 20-Jun-2015

533 Views

Category:

Technology

0 Downloads

Preview:

Click to see full reader

DESCRIPTION

http://ceur-ws.org/Vol-902/paper_7.pdf Knowledge on the Web comes in ever larger amounts and in a wider variety of structure and semantics that ever before. In or- der to exploit this knowledge in di erent applications, many researchers investigate techniques for making sense of Web data. Objects that the techniques try to identify and extract are, for example, people, organiza- tions, and locations. Many applications though observe how events play an increasingly more important role. Capturing and extracting events for sense making analysis is what this research is aiming for, and in this paper we present the rst results and contributions from our research. We consider how events get extracted, how they get conceptualized, and how visual analytics helps to make sense of the represented events. All of this is illustrated in a representative example where driven by questions from social scientists we apply our pipeline to the domain of activism, e.g. Occupy, Arab Revolution.

TRANSCRIPT

Making Sense of the Arab Revolution & OccupyExtracting, Aggregating & Visualizing Events

Thomas Ploeger, Bibiana Armenta, Lora Aroyo, Frank de Bakker, Iina Hellsten

Monday, November 12, 12

When we talk about events we want to know ...

• Which activists were most active in 2011?

• Which Dutch activist organizations have been involved in labour strike events?

• What type of activists events happened in Berlin last year?

• What is the most popular activist event so far?

Monday, November 12, 12

When we talk about events we want to know ...

• Which activists were most active in 2011?

• Which Dutch activist organizations have been involved in labour strike events?

• What type of activists events happened in Berlin last year?

• What is the most popular activist event so far?

Monday, November 12, 12

Types of queries• Descriptive

• What are the properties of the networks that are formed around a campaign?

• Which tactics activist groups apply most often to influence norms of companies on issues of CSR?

• Narrative

• Which actors use a specific tactic at a specific point in time?

• Interpretative

• How does non-traditional media influence activist groups' tactics and positions (blogs, social media in general)?

• Why is one form of tactic chosen by an activist group rather than another one?

• Is it related to the campaign, or to the targeted company

• Is it related to the time, technology, place, or culture?

• Is it related to the tradition of this group?

Monday, November 12, 12

Types of queries• Descriptive

• What are the properties of the networks that are formed around a campaign?

• Which tactics activist groups apply most often to influence norms of companies on issues of CSR?

• Narrative

• Which actors use a specific tactic at a specific point in time?

• Interpretative

• How does non-traditional media influence activist groups' tactics and positions (blogs, social media in general)?

• Why is one form of tactic chosen by an activist group rather than another one?

• Is it related to the campaign, or to the targeted company

• Is it related to the time, technology, place, or culture?

• Is it related to the tradition of this group?

Monday, November 12, 12

Occupy Wall Street

a police officer sprayed 4 protesters with pepper spray

• Protesters argued that the use of pepper spray was uncalled for vs. necessary (NYPD defended the officer)

• The officer stated that the event was taken out of context vs. at fault (investigation concludes that the officer was at fault and he was reprimanded)

Monday, November 12, 12

Self-immolation of MB

Mohamed Bouazizi, Tunisian street vendor set himself on fire to protest after officials confiscated his wares.

• personal, economic motivation vs. martyr

• the spark that ignited Tunisian Revolution & Arab Spring vs. singular personal event

• How was he treated by officials when they confiscated his wares?

• How did officials respond to his complaints?

• Are there any earlier encounters between him and officials?

Monday, November 12, 12

Why is it difficult?• Information is

• scattered across different sources

• offered in different formats

• often incomplete, incorrect, out of context, bias

• Events are

• perceived from different points of view

Monday, November 12, 12

alleviating bias by modeling & visualizing activist

Mapping Online Networks of Activism

Monday, November 12, 12

Activist Events Terminology

• Activist Event: an action undertaken by an actor as part of a campaign with the aim of influencing the state (e.g. resolved) of an issue

• Tactic: indicates the event type

• Actor: a person, group or organization of a given type (e.g. radical, reformative) performing tactics

• Company: an organization that triggers an issue

• Issue: a topic or problem important to actors and companies

• Campaign: consists of a set of events undertaken by an actor aiming to influence the state of an issue

Monday, November 12, 12

Modeling Activist Events

Monday, November 12, 12

Modeling Activist Events

campaign-centered

Monday, November 12, 12

Modeling Activist Events

campaign-centered

event-centered

Monday, November 12, 12

ACTEVE-SEMACTivist EVEnts model based on Simple Event Model

• actors, roles, objects, places, times

• viewpoints (according to a certain authority), i.e.attribution of authoritative source of a statementtime-bounded validity of factsevent-bounded roles

Monday, November 12, 12

Example I

MultipleSources

MultipleRepresentations

All versions shownvisually, comparatively

Event that is reported differently by multiple sources

Monday, November 12, 12

Example I

MultipleSources

MultipleRepresentations

All versions shownvisually, comparatively

Event that is reported differently by multiple sources

event-bounded roles of actors, e.g. police officer - aggressor vs. peacekeepersem:View sem:Authority

Monday, November 12, 12

Example IIEvent that is reported with or without incorrect context

SourceMaterial

EarlierEvents

EventRepresentation

Event shown in contextof earlier events

Monday, November 12, 12

Example IIEvent that is reported with or without incorrect context

SourceMaterial

EarlierEvents

EventRepresentation

Event shown in contextof earlier events

time-bounded validity of facts, e.g. Mr. Bouazizi - street vendor vs. martyrsem:Temporary

Monday, November 12, 12

Candidate Events & Entities

Monday, November 12, 12

Candidate Events & Entities

Monday, November 12, 12

Timelines from Text

4 right, 2 wrong, 3 missing eventstwo have no explicit times & are in the wrong order

One involved al-Qaeda but took place in Jordan on the Syrian border

does a fuzzy task require fuzzy metrics?

“al-Qaeda activities in Syria”

set of events partially ordered by time, e.g. before/after

Monday, November 12, 12

Metrics• how to measure partial & overall correctness,

• compared to the worst/optimal/best timeline

• how timeline should optimize for missing data, e.g. times, locations (deduce temporal and spatial position)

• in context of the task or purpose it is used for

• for a given type of queries

• how to determine importance of events (also in context)

• how to determine importance of dimensions (also in context)

• are there dependencies between the different dimensions

• measuring overall coherence of timeline

Monday, November 12, 12

Monday, November 12, 12

Monday, November 12, 12

When visualizing ...• Adjusting the granularity in

terms of times, locations and event types

• Comparing different perspectives on the same event

• Showing evolution over time of event types, event participants or location

• Filtering events based on (strength of) connections to other events, participants, places or periods

Monday, November 12, 12

Monday, November 12, 12

Monday, November 12, 12

Monday, November 12, 12

Monday, November 12, 12

Monday, November 12, 12

Monday, November 12, 12

Questions?

@laroyohttp://lora-aroyo.org

Monday, November 12, 12

top related