The Social Semantics of LiveJournal FOAF: Structure and Change from 2004 to 2005 - PowerPoint PPT Presentation

1 / 33
About This Presentation
Title:

The Social Semantics of LiveJournal FOAF: Structure and Change from 2004 to 2005

Description:

Entries often have hyperlinks, especially to other weblogs ... David Bowie, Radiohead, Pink Floyd, Led Zeppelin, Ani DiFranco, Modest Mouse, etc. ... – PowerPoint PPT presentation

Number of Views:41
Avg rating:3.0/5.0
Slides: 34
Provided by: blogn5
Category:

less

Transcript and Presenter's Notes

Title: The Social Semantics of LiveJournal FOAF: Structure and Change from 2004 to 2005


1
The Social Semantics of LiveJournal FOAF
Structure and Change from 2004 to 2005
  • ISWC2005 Workshop on
  • Semantic Network AnalysisNovember 7, 2005

2
Weblogs and Online Journals
  • Weblog a regularly updated web page, with
    entries in reverse chronological order
  • Facilitated by blogging software
  • Hosted on many central sites
  • Entries often have hyperlinks, especially to
    other weblogs
  • Online journal a regularly updated web page
    hosted on a site that provides facilities for
    managing social contacts
  • blog.livedoor.com
  • www.livejournal.com
  • www.mypress.com
  • www.deadjournal.com
  • www.xanga.com
  • yaplog.jp

3
(No Transcript)
4
(No Transcript)
5
(No Transcript)
6
(No Transcript)
7
(No Transcript)
8
Social Network Metadata
  • Friend-of-a-Friend (FOAF) project
  • Initiated by World-Wide Web Consortium
  • Dan Brickley, Libby Miller, John Breslin, Stefan
    Decker, etc.
  • 1997 to present
  • A practical experiment
  • Use semantic web to make (virtual) communities
    self-aware
  • Target social networking applications
  • tribe.net, ecademy, Orkut, etc.

9
(No Transcript)
10
Problems
  • What sort of relationships typically exist within
    the data provided as FOAF by LiveJournal?
  • - when relationships do exist, how stable are
    they?
  • Does it contain discernable and useful patterns?
  • What sorts of inference might such patterns
    support?

11
Method
  • Specifically focus on
  • foafinterests
  • foafknows
  • Quantitative analysis of a substantial sample of
    FOAF data
  • Principal Components Analysis
  • Hierarchical clustering
  • Social network analysis
  • Comparison at two different points in time
  • Examine patterns for stability

12
The Initial Sample from some of our previous
work.
  • User profiles crawled over 5 days, March 2004
  • Courtesy of Jim Ley
  • Total of 33512 LiveJournal FOAF profiles.
  • http//www.jibbering.com/foaf/dumps/
  • This produced a sample of convenience, not a true
    random sample. However, it does produce a cross
    section of LiveJournal users which represent the
    diversity of the LJ users as a whole.

13
Interests
  • Incidence matrix
  • 500 most common interests, 11K participants
  • normalized (z-score)
  • Principal components analysis
  • 20 dimensions selected for interpretation
  • Component scores for all participants
  • Component loadings for all 500 interests
  • Hierarchical cluster analysis
  • on participants groups for SNA
  • on interests for interpretation of social groups
  • Cuts at 9 and 76 groups

14
Interest Clusters
  • Red General (intellectual pursuits)
  • Orange Art
  • Lt Green Nightlife
  • Green Romance
  • Cyan Sexuality
  • Lt Blue Grunge music, piercing/tattoos
  • Dk Blue Sci-fi fantasy comics
  • Violet Music
  • Magenta Goth/fetish

15
Red General/ Intellectual Pursuits
Christianity
outdoors/ nature
culture centers
leisure
Philosophy
Feminism
home
kids
cinema
science
arts
drama
16
Cyan Sexuality
ambisexual
men/women
passion
pornography
17
Magenta Goth Fetish
Industrial
Fetish
Goth
18
Interests
  • Principal Components

19
Niche
Mainstream
High Freq
Low Freq
20
Intellectual
Sensual
Society
Fantasy
21
Experience
Expression
Social
Private
22
???
???
Pagan
Christian
23
Stability
  • Are the patterns stable over time?

24
A (Partly) New Sample
  • 2004 sample
  • 500 most popular interests
  • 500 most popular users
  • 2005 re-sample
  • Same set of users as before, same set of
    interests and popular users
  • Combined incidence matrix
  • PCA (z-score column-wise normalized)
  • Hierarchical cluster analysis (separate by
    relation and year)

25
knows
interest
26
(No Transcript)
27
2005
2004
28
  • Five Cluster Cut by Year
  • 2004 interests
  • Science Fiction, Fantasy, Celtic, and graphic
    arts,
  • General interests
  • Sex, goth subculture, body modification and
    fetish
  • A variety of contemporary music interests (The
    Cure, Joy Division,
  • David Bowie, Radiohead, Pink Floyd, Led
    Zeppelin, Ani DiFranco,
  • Modest Mouse, etc.)
  • Industrial and alternative rock music.
  • 2005 interests
  • Social, natural, and mystical interests
  • General interests
  • Ska, punk and alternative rock music
  • Science Fiction, Fantasy, Celtic, and graphic
    arts
  • Sex, goth subculture, body modification and
    fetish

29
Global Change 2004-2005
foafknows
foafinterests
30
Summary Observations
  • In both 2004 and 2005
  • foafinterests and foafknows are uncorrelated
    with each other
  • foafinterests and foafknows are strongly
    intra-correlated
  • correlations of interests are weaker than those
    of knows
  • Local groupings show a great deal of change
  • strong inferences are not licensed
  • Global change from 2004 to 2005 is minimal
  • slight weakening of most popular 2004 interests
  • slight strengthening of most popular friends
    (celebs)

31
Remaining Issues
  • Sampling issues
  • The scale of the available data is very large
  • Snowball sampling from the large site makes it
    unclear what piece of the network we get
  • Additional random sampling points may improve
    this
  • Dynamics
  • We need to know more about how the interests and
    friends relations change over time

32
References
  • Degenne, Alain and Michel Forse. 1999.
    Introducing Social Networks. Thousand Oaks,
    California Sage Publications.
  • FOAF Vocabulary Specification. http//xmlns.com/fo
    af/0.1/ Accessed July 18, 2004.
  • Granovetter, M. The strength of weak ties.
    American Journal of Sociology, 78(6), 1360-1380
    (1973).
  • Forse, Michel. 1981. La sociabilite. Economie et
    Statistique 13239-48.
  • Kling, Rob. What is Social Informatics and Why
    Does it Matter? D-Lib Magazine. Volume 5 Number 1
    (January 1999).
  • Lavoie, Brian et al. Trends in the Evolution of
    the Public Web. D-lib Magazine v9.4, April 2003.
    Available at http//www.dlib.org/dlib/april03/lavo
    ie/04lavoie.html. Accessed July 18, 2004
  • O'Neill, Edward et al. A Methodology for Sampling
    the World Wide Web. Available at
    http//digitalarchive.oclc.org/da/ViewObject.jsp?o
    bjid0000003447. Accessed July 18, 2004.
  • Paolillo, John C. 2001. Language variation on
    Internet Relay Chat A social network approach.
    Journal of Sociolinguistics 5180-213.
  • Reid, E. M. 1991. Electropolis Communication and
    Community on Internet Relay Chat. B.A. Honours
    Thesis, Department of History, University of
    Melbourne.
  • RDF Syntax Specification. http//www.w3.org/TR/rdf
    -syntax-grammar/. Accessed July 18, 2004.
  • W3C Semantic Web Activity Statement.
    http//www.w3.org/2001/sw/Activity. Accessed July
    18, 2004.
  • Wasserman, Stanley and Katherine Faust. Social
    Network Analysis. Cambridge Cambridge UP, 1994.

33
THE END
Write a Comment
User Comments (0)
About PowerShow.com