STUDY the Social Web through Analysis and Network Science
1. Social WebāØ
2016
Lecture 6: How can we STUDY the Social Web?
(based on slides from Les Carr, Nigel Shadbolt, Harith Alani)
Davide Ceolin (credits to: Lora Aroyo)
The Network Institute
VU University Amsterdam
2. The Web
the most used and one of the most transformative applications
in the history of computing, e.g. how the Social Web has
transformed the world's communication
approximately 1010 people
more than 1011 web documents
Social Web 2016, Davide Ceolin
3. The Web
Great success as a technology,
itās built on signiļ¬cant computing infrastructure,
but
as an entity surprisingly unstudied
Social Web 2016, Davide Ceolin
4. ā¢ physical science: analytic discipline to ļ¬nd laws that
generate or explain observed phenomena
ā¢ CS is mainly synthetic: formalisms & algorithms are
created to support speciļ¬c desired behaviors
ā¢ Web Science: web needs to be studied & understood
as a phenomenon but also to be engineered for future
growth and capabilities
Science & Engineering
Social Web 2016, Davide Ceolin
8. Web is NOT a Thing
ā¢ itās not a verb, nor a
noun
ā¢ itās a performance, not
an object
ā¢ co-constructed with
society
ā¢ activity of individuals
who create interlinked
content that reļ¬ect &
reinforce the
interlinkedness of society
& social interaction
... and a record of
that performance
Social Web 2016, Davide Ceolin
10. eScience: Analysis of Data
ā¢ the automated or semi-automated extraction of
knowledge from massive volumes of data ā it is a
lot, but it is not just a matter of volume
ā¢ 3 Vs of Big Data
ā¢ Volume: # rows / object / bytes
ā¢ Variety: # columns / dimensions / sources
ā¢ Velocity: # columns / bytes per unit time
ā¢ more Vs ā Veracity: Can we trust this data?
Social Web 2016, Davide Ceolin
11. Simple micro rules give rise to
complex macro phenomena
ā¢ at microscale an infrastructure of artiļ¬cial languages and protocols:
a piece of engineering
ā¢ however, interaction of people creating, linking and consuming
information generates web's behavior as emergent properties at
macroscale
ā¢ properties require new analytic methods to be understood
ā¢ some properties are desirable and are to be engineered in, others
are undesirable and if possible engineered out
Social Web 2016, Davide Ceolin
12.
13. ā¢ software applications designed based on appropriate
technology (algorithm, design) and with envisioned
'social' construct
ā¢ usually tested in the small, testing microscale properties
ā¢ a macrosystem evolving from people using the
microsystem and interacting in often unpredicted ways, is
far more interesting and must be analyzed in different
ways
ā¢ macrosystems exhibit challenges that do not exist at
microscale
A new way of software
development
Social Web 2016, Davide Ceolin
14. Example:
Evolution of Search Engines
1: techniques designed to rank documents
2: people were gaming to inļ¬uence algorithms &
improve their search rank
3: adapt search technologies to defeat this inļ¬uence
Social Web 2016, Davide Ceolin
15. Web Science Reļ¬ections
Is the Web changing faster than our ability to observe it?
How to measure or instrument the Web?
Social Web 2016, Davide Ceolin
16. The Web Graph
ā¢ to understand the web, in good CS
tradition, we look at the graph
ā¢ nodes are web pages (HTML)
ā¢ edges are hypertext links
between nodes
ā¢ ļ¬rst analysis shows that in-degree
and out-degree follow power law
distribution => holds for large
samples
ā¢ this gave insight into the growth of
the web
Social Web 2016, Davide Ceolin
17. The (Search) Algorithms
ā¢ the Web graph also as basis of
algorithms for search engines:
ā¢ PageRank and others
assume that inserting a
hyperlink symbolizes an
endorsement of authority of
the page linked to
Social Web 2016, Davide Ceolin
18. User State is Important
ā¢ the original Web graph is too simple, starts from quasi static HTML
ā¢ for personalization or customization different representations (of
sources) may be served to different requesters, e.g. cookies
ā¢ graph-based models often do not account for this sort of user-
dependent state, and not ļ¬t for all the information behind the
servers, in DeepWeb
ā¢ itās not a simple HTTP-GET anymore (but HTTP-POST or HTTP-
GET with complex URI) that is the basis for deļ¬ning nodes in the
graph
ā¢ URis that carry user state are heavily used in Web applications, but
are not in the model and largely unanalyzed
Social Web 2016, Davide Ceolin
19. According to Google
each day 20-25% of searches have not been seen before, i.e.
generate a new identiļ¬er
thus a new node in the graph
more than 20 million new links per day, 200 per second
do they follow the same power laws & growth models?
validating such models is hard
exponential growth of content
changes in number & power of servers
increasing diversity in users
Social Web 2016, Davide Ceolin
20. Wikipedia
ā¢ purely mathematical (technology-based) models do not capture the
whole story
ā¢ the Wikipedia structure (link labels) shows a Zipf-like distribution
just like other tag-based systems
ā¢ Wikipedia is built on MediaWiki software
ā¢ but other MediaWiki-based applications did not generate such
signiļ¬cant use
ā¢ the pure 'technological' explanation cannot explain it
ā¢ must be related to the 'social model' of how Wikipedia is
organized
this is referred to as the dynamics of a 'social machine' (already inTBLās original vision ofWWW)
Social Web 2014, Lora Aroyo!
21. Collective Intelligence
ā¢ why do people contribute?
ā¢ how to maintain the connected content?
ā¢ how are trust & provenance represented, maintained
and repaired on the Web?
Social Web 2014, Lora Aroyo!Social Web 2016, Davide Ceolin
22. Collective Intelligence
Motivation Example Mean
Fun āWriting inWikipedia is funā 6,1
Ideology āI think information should be freeā 5,59
Values āI feel itās important to help othersā 3,96
Understanding āWriting inWikipedia allows me to gain a new perspective on thingsā 3,92
Enhancement āWriting inWikipedia makes me feel neededā 2,97
Protective āBy writing inWikipedia I feel less lonelyā 1,97
Career āI can make new contacts that might help my careerā 1,67
Social āPeople I am close to want me to write inWikipediaā 1,51
Social Web 2016, Davide Ceolin
23. Social Machines
ā¢ today's interactive applications are very early
social machines limited by being largely isolated from
one another
ā¢ more effective social machines can be expected
ā¢ social processes in society interlink, so they
should also interlink on the web
ā¢ technology needed to allow user communities to
construct, share & adapt social machines to get
success through trial, use & reļ¬nement
Social Web 2014, Lora Aroyo!Social Web 2016, Davide Ceolin
25. itās relationships, stupid!
not attributes
May, 2007April, 2002
All the world's a net
by David Cohen
Social Web 2016, Davide Ceolin
26. Think Networks!
ā¢ everything is connected to everything else
ā¢ networks are pervasive - from the human brain
to the Internet to the economy to our group of
friends
ā¢ following underlying order and follow simple laws
ā¢ "new cartographers" are mapping networks in a
wide range of scientiļ¬c disciplines
ā¢ social networks, corporations, and cells are more
similar than they are different
ā¢ new insights into the interconnected world
ā¢ new insights on robustness of the Internet, spread
of fads and viruses, even the future of democracy.
Albert-LĆ”szlĆ³ BarabĆ”si: Linked:The New Science of Networks
April, 2002
Social Web 2016, Davide Ceolin
28. Networks:
another perspective
ā¢ Social Networks: Itās not what you know,
itās who you know
ā¢ Cognitive Social Networks: Itās not who
you know, itās who they think you know.
ā¢ Knowledge Networks: Itās not what you
know, itās what they think you know
Social Web 2016, Davide Ceolin
29. Network
Analysis
ā¢ is about linking social actors, e.g.
systematically understanding
and identifying connections
ā¢ by using empirical data
ā¢ draws on graphic imagery
ā¢ relies on mathematical/
computational models
ā¢ Jacob Moreno - one of the
founders of social network
analysis; some of the earliest
graphical depictions of social
networks (1933)
Social Web 2016, Davide Ceolin
30. Leveraging recent advances in:
ā¢ Theories: about social motivations for creating, maintaining, dissolving & re-creating
links in multidimensional networks & about emergence of macro-structures
ā¢ Data: Semantic Web provides technological capability to capture, store, merge &
query relational metadata to more effectively understand & enable communities
ā¢ Methods: qualitative & quantitative for theoretically-grounded network predictions
ā¢ Computational infrastructure: Cloud computing & petascale applications are
critical to face the computational challenges in analyzing the data
Social Web 2016, Davide Ceolin
32. Web Science is about
additionality
not the union of
disciplines, but
intersection
Social Web 2016, Davide Ceolin
33. Society is Diverse
different parts of society have different objectives and hence incompatible
Web requirements, e.g. openness, security, transparency, privacy
Social Web 2016, Davide Ceolin
34. ā¢ POWER DISTANCE:The extent to which power
is distributed equally within a society and the
degree that society accepts this distribution.
ā¢ UNCERTAINTY AVOIDANCE:The degree to
which individuals require set boundaries and
clear structures
ā¢ INDIVIDUALISM vs COLLECTIVISM:The degree
to which individuals base their actions on self-
interest versus the interests of the group.
ā¢ MASCULINITY vs FEMININITY:A measure of a
society's goal orientation
ā¢ TIME ORIENTATION:The degree to which a
society does or does not value long-term
commitments and respect for tradition.
Understanding the
Socio-Cultural
Social Web 2016, Davide Ceolin
35. Understanding variations
ā¢ Ecology of theWeb - structure of
the environment, producers
and consumers
ā¢ Populations (individuals and
species), traits/characteristics,
heredity, genotypes and
phenotypes
ā¢ Mechanisms - variation
(mutation, migration, genetic
drift), selection
ā¢ Outcomes - adaption, co-
evolution, competition, co-
operation, speciation, extinction
Social Web 2016, Davide Ceolin
36. but
How to do the Science?
Social Web 2016, Davide Ceolin
37. Big Data Owners
Who can do macro analysis?
ā¢ Google, Bing,Yahoo!, Baidu
ā¢ Large scale, comprehensive data
ā¢ New forms of research alliance
How Billions ofTrivial Data Points can Lead to
Understanding
Social Web 2016, Davide Ceolin
40. Web Science Reļ¬ections
How to identify behaviors and patterns?
How to analyze the changing structure of the Web?
Social Web 2016, Davide Ceolin
41. The Age of OPEN Data
Social Web 2016, Davide Ceolin
42. The Age of OPEN Data
TRANSPARENCY VALUE ENGAGEMENT
ā¢ common standards for release of public data
ā¢ common terms for data where necessary
ā¢ licenses - CC variants
ā¢ exploitation & publication of distributed, decentralised information assets
Social Web 2016, Davide Ceolin
44. Big Bang:
Web Information
ā¢ the assumption of open exchange of information is
being imposed on the society
ā¢ is the Web, and its open access, open data, scientiļ¬c &
creative commons offer a beneļ¬cial opportunity or
dangerous cul-de-sac?
Social Web 2016, Davide Ceolin
45. Open Questions
ā¢ How is the world changing as other parts of society impose their
requirements on the Web?, e.g. current examples with SOTA/PIPA,ACTA
requirements for security and policing taking over free exchange of information,
unrestricted transfer of knowledge
ā¢ Are the public and open aspects of the Web a fundamental change in
societyās information processes, or just a temporary glitch?, e.g. are open
source, open access, open science & creative commons efļ¬cient alternatives to
free-based knowledge transfer?
Social Web 2016, Davide Ceolin
46. Open Questions
ā¢ do we take Web for granted as provider of a free & unrestricted
information exchange?
ā¢ is Web Science the response to the pressure for the Web to change - to
respond to the issues of security, commerce, criminality & privacy?
ā¢ what is the challenge for Web science in explaining how the Web impacts
society?
Social Web 2016, Davide Ceolin
47. What can you do as a
Computer Scientist?
speciļ¬cally for the SocialWeb
Social Web 2016, Davide Ceolin
48. Hands-on Teaser
ā¢ Present your social web app pitch
ā¢ 12 March (10:00 - 12:45)
ā¢ C.623 all groups together
ā¢ 1 mins presentation time
ā¢ be on time
ā¢ send your slide(s) the day before via the website
Social Web 2016, Davide Ceolin