crowdsourcing transparency requirements through structured feedback and social
TRANSCRIPT
1
Crowdsourcing Transparency Requirements through Structured Feedback and Social Adaptation
Mahmood Hosseini, Alimohammad Shahri, Keith Phalp and Raian AliBournemouth University
01/06/2016
2
Agenda Introduction
What are transparency requirements? Peculiarities of transparency
Motivation Why crowdsourcing? Why structured feedback? Why social adaptation?
Overviews Overview of transparency reference models Overview of our proposed framework Overview of structured feedback elements
Applying the concepts in the engineering of transparency requirements Applying crowdsourcing Applying structured feedback Applying social adaptation
Conclusion and future work
01/06/2016
Intro
duct
ion
Mot
ivat
ion
Over
view
sAp
plica
tions
Conc
lusio
n
3
Transparency RequirementsIn
trodu
cti
onM
otiv
atio
nOv
ervi
ews
Appl
icatio
nsCo
nclu
sion
What information should I put on my product label?
(THE RIGHT TO SHOW)
01/06/2016
4
Transparency RequirementsIn
trodu
cti
onM
otiv
atio
nOv
ervi
ews
Appl
icatio
nsCo
nclu
sion
What information do I need to know in a T&C?
(THE RIGHT TO KNOW)
01/06/2016
5
Transparency Requirements Non functional requirements Getting more attention
The change of the millennium Social crises, e.g., Panama Papers and Ashley
Madison Financial crises, e.g., the one in 2008
01/06/2016
Intro
duct
ion
Mot
ivat
ion
Over
view
sAp
plica
tions
Conc
lusio
n
Image courtesy of Time, 14 Apr 2016
6
Transparency Requirements Initiatives to handle transparency
requirements E.g., Freedom of Information Act, Open
Government Need for engineering
Reference models, conceptual models, tools, automated analysis
01/06/2016
Intro
duct
ion
Mot
ivat
ion
Over
view
sAp
plica
tions
Conc
lusio
n
7
Peculiarities of Transparency What to disclose
Informational needs of the stakeholders Who to disclose to
Stakeholder identification, their different needs When to disclose
Timeliness How much to disclose
Preventing information overload Regulations and limitations
Other non-functional requirements such as privacy Evolution over time
Information life span, transient requirement Targeted transparency vs. tailored transparency
Role level vs. individual level01/06/2016
Intro
duct
ion
Mot
ivat
ion
Over
view
sAp
plica
tions
Conc
lusio
n
8
But what is transparency anyway?
Transparency is
providing high
quality informatio
n
Transparency should be meaningful for its users
Transparency is
synonymous with
information
Transparency
should only
reach those who
need it
01/06/2016
Intro
duct
ion
Mot
ivat
ion
Over
view
sAp
plica
tions
Conc
lusio
n
Transparency is about accessible,
understandable
information
9
Motivation:Why Crowdsourcing? Crowdsourcing allows the
engagement of a large, diverse crowd
It has been investigated in RE activities StakeRare, StakeSource, REfine,
CrowdREquire
It can be considered as a solution to dynamic or bi-directional transparency
01/06/2016
Intro
duct
ion
Mot
ivat
ion
Over
view
sAp
plica
tions
Conc
lusio
n
10
Motivation:Why Structured Feedback? Easier to aggregate, process,
analyse and evaluate
Its use has been investigated in crowdsourcing activities
Transparency attributes allow for structured feedback
01/06/2016
Intro
duct
ion
Mot
ivat
ion
Over
view
sAp
plica
tions
Conc
lusio
n
11
Motivation:Why Social Adaptation? Social adaptation regards user
feedback as the main driving force for software evolution
A good fit for transparency requirements They are volatile They change over time They vary for different stakeholders
We need social adaptation to ensure the crowd is heard01/06/2016
Intro
duct
ion
Mot
ivat
ion
Over
view
sAp
plica
tions
Conc
lusio
n
12
Overview of Transparency Reference Models
01/06/2016
Intro
duct
ion
Mot
ivat
ion
Over
view
sAp
plica
tions
Conc
lusio
n
(Presented in RE 2015)
13
Overview of Transparency Reference Models
01/06/2016
Intro
duct
ion
Mot
ivat
ion
Over
view
sAp
plica
tions
Conc
lusio
n
(Facets Presented in REFSQ 2016)
14
Overview of Transparency Reference Models
01/06/2016
Intro
duct
ion
Mot
ivat
ion
Over
view
sAp
plica
tions
Conc
lusio
n
(Facets Presented in REFSQ 2016)
15
Overview of Transparency Reference Models
01/06/2016Kahn, B. K., Strong, D. M., & Wang, R. Y. (2002). Information quality benchmarks: product and service performance. Communications of the ACM, 45(4), 184-192.
Intro
duct
ion
Mot
ivat
ion
Over
view
sAp
plica
tions
Conc
lusio
n
(Fac
ets P
rese
nted
in R
EFSQ
201
6)
16
Overview of Our Proposed Framework
01/06/2016
Intro
duct
ion
Mot
ivat
ion
Over
view
sAp
plica
tions
Conc
lusio
n
monitor
analyse planexecute
17
Overview of Structured Feedback Elements
Intro
duct
ion
Mot
ivat
ion
Over
view
sAp
plica
tions
Conc
lusio
n
01/06/2016
18
Applying Crowdsourcing in the Engineeringof Transparency Requirements The four pillars of
crowdsourcing match transparency actors The crowd: information
receivers The crowdsourcer: information
provider The crowdsourced task:
transparency provision The crowdsourcing platform:
information medium01/06/2016
Intro
duct
ion
Mot
ivat
ion
Over
view
sAp
plica
tions
Conc
lusio
n
19
Applying Crowdsourcing in the Engineeringof Transparency Requirements Crowdsourcing for the identification
of stakeholders Identification of alternative information
providers, as acknowledged in Section 2(21)(1) of UK FOIA
Identification of alternative information mediums
Creation of alternative information mediums, e.g., by sharing information in social media
Identification of other information receivers, i.e., spreading transparency to those who need it
01/06/2016
Intro
duct
ion
Mot
ivat
ion
Over
view
sAp
plica
tions
Conc
lusio
n
20
Applying Crowdsourcing in the Engineeringof Transparency Requirements Spotting transparency
meaningfulness mismatch (i.e., mismatches in data, process, and/or policy transparency) Annotation of information as data, process and
policy Notifying where information concerning data,
process, and policy is missing through requests of information
01/06/2016
Intro
duct
ion
Mot
ivat
ion
Over
view
sAp
plica
tions
Conc
lusio
n
21
Applying Crowdsourcing in the Engineeringof Transparency Requirements Information availability:
Missing information Disseminating information for better
reach
Information interpretation: Crowd interpretation of the
information into a more crowd-friendly language
Helping formal interpreters, e.g., news agencies in local issues
Compare interpretations01/06/2016
Intro
duct
ion
Mot
ivat
ion
Over
view
sAp
plica
tions
Conc
lusio
n
22
Applying Crowdsourcing in the Engineeringof Transparency Requirements Information accessibility:
Spotting difficulties in access, e.g., too many clicks, obsolete electronic formats
Enhancing accessibility by sharing
Information perception: Perception sharing to help
information providers reduce the perception gap
01/06/2016
Intro
duct
ion
Mot
ivat
ion
Over
view
sAp
plica
tions
Conc
lusio
n
23
Applying Crowdsourcing in the Engineeringof Transparency Requirements Information understandability:
Highlighting ambiguities, discrepancies, etc., e.g., in a Terms and Conditions document
Information acceptance: Sharing their acceptance/refusal,
and the reason for it, e.g., lack of reputation of the news source
01/06/2016
Intro
duct
ion
Mot
ivat
ion
Over
view
sAp
plica
tions
Conc
lusio
n
24
Applying Crowdsourcing in the Engineeringof Transparency Requirements Information actionability:
Sharing if and how information is useful, similar to celebrity endorsements, e.g., an increase in the annual fee resulting in bank customers moving their money to other banks
01/06/2016
Intro
duct
ion
Mot
ivat
ion
Over
view
sAp
plica
tions
Conc
lusio
n
25
Applying Crowdsourcing in the Engineeringof Transparency Requirements Product/service quality
dimensions conforming to specifications Finding inconsistencies, imprecisions,
etc.
Product/service quality dimensions meeting or exceeding consumer expectations Crowd feedback must be elicited in
the quality assurance of these dimensions
01/06/2016
Intro
duct
ion
Mot
ivat
ion
Over
view
sAp
plica
tions
Conc
lusio
n
26
Applying Structured Feedback in the Engineering of Transparency Requirements Pinpointing the loci where transparency
has been successful or has failed: On the level of transparency required (i.e.,
data, process, policy) On the last step where transparency has
been achieved or failed (i.e., availability, interpretation, accessibility, perception, understandability, acceptance, actionability)
On the information quality in transparency On stakeholder identification and medium
discovery Still, the use of free-form comment is
recommended along with structured feedback
01/06/2016
Intro
duct
ion
Mot
ivat
ion
Over
view
sAp
plica
tions
Conc
lusio
n
27
Applying Social Adaptation in the Engineering of Transparency Requirements Transparency requirements are often
individual, context-dependent, and emerging requirements
Every individual’s transparency requirement should be ideally elicited, similar to the industry idea of mass customisation
For transparency, a continuous feedback acquisition facilitates mass customisation How, what, where, when to disclose and to whom
01/06/2016
Intro
duct
ion
Mot
ivat
ion
Over
view
sAp
plica
tions
Conc
lusio
n
28
Applying Social Adaptation in the Engineering of Transparency Requirements Monitoring:
Stakeholders’ transparency requirements change with the change of time and context, and they often become obsolete as soon as they are met
Monitoring through feedback is necessary to ensure transparency requirements are met all the time
The power of the crowd can be harnessed through structured feedback for the engineering of transparency requirements
01/06/2016
Intro
duct
ion
Mot
ivat
ion
Over
view
sAp
plica
tions
Conc
lusio
n
29
Applying Social Adaptation in the Engineering of Transparency Requirements Analysis:
Structured feedback can help the analysis of transparency requirements
Modelling and formalisation of transparency requirements helps reasoning and analysis
We have proposed a domain-specific modelling language, TranspLan, for the engineering of transparency requirements ▪ (to be presented in CAiSE 2016)
01/06/2016
Intro
duct
ion
Mot
ivat
ion
Over
view
sAp
plica
tions
Conc
lusio
n
30
Applying Social Adaptation in the Engineering of Transparency Requirements Planning:
Finding the best alternative for satisficing transparency requirements
Using recommender systems helps in the selection of alternatives
User profiling helps recommender systems understand users’ interests
Execution: Just consider that transparency
requirements are transient!01/06/2016
Intro
duct
ion
Mot
ivat
ion
Over
view
sAp
plica
tions
Conc
lusio
n
31
Conclusion and Future Work We proposed a conceptual
framework for utilising crowdsourcing and social adaptation through structured feedback for engineering and evolving transparency requirements
Our future work will be consolidating the proposed framework, and providing methods supported by automated tools for covering the entire life cycle of transparency requirements
01/06/2016
Intro
duct
ion
Mot
ivat
ion
Over
view
sAp
plica
tions
Conc
lusio
n
monitor
analyse
plan execute
32
Acknowledgements
The research is supported by an FP7 Marie Curie CIG grant (the SOCIAD project).
01/06/2016
33
Thank You!
01/06/2016