Generating and Tracking Communities Based on Implicit Affinities

1
Generating and Tracking CommunitiesBased on
Implicit Affinities

• Matthew Smith (smitty_at_byu.edu)
• Christophe Giraud-Carrier and Brock Judkins
• BYU Data Mining Lab
• Spring Research Conference 2007

2
Outline

• Introduction
• Community Generation IANs
• Social Capital for Community Tracking
• Experiments Observations
• Conclusions
• Future Work

3
Introduction

• Online Communities
• Continually emerging many sites is adding this
  aspect
• Like offline communities, they are complex and
  dynamic
• Blogosphere focus of experiments
• Generation
• Individuals as social actors characterized by
  attributes
• Unlike traditional social networks where links
  represent explicit relationships, the links in
  our approach are based strictly on affinities
• We let relationships among them emerge naturally
• Tracking
• Explore the structure and behavior of communities

4
Community Generation IANs
5
Implicit Affinity

• Affinity
• The overlapping of attributes-values for any
  common attribute
• Community
• An arbitrary number of individuals characterized
  by attributes
• Linked by affinities rather than explicit
  relationships

6
Affinity Scoring

• Affinity score for a particular attribute
• Affinity score for all attributes

7
Affinity Network Building
IAN
8
Social Capital for Community Tracking
9
Social Capital Bridging and Bonding

• Social Capital The advantage available through
  connections between individuals within a
  particular network
• Bonding and Bridging Metrics

10
Experiments Observations
11
Scobleizers Blog List

• Robert Scoble (Scobleizer)
• Blogger and book author
• Technical evangelist (formerly with Microsoft)
• Data Set Details
• Scobleizers reading list at Bloglines.com
• 570 blogs
• 2380 bloggers

12
Blog posts per day
Lack of data for all bloggers during first few
days
We observe fewer posts during the weekend (Friday
Saturday)
13
Single Attribute Companies

• Motivation
• Many bloggers talk about various companies and
  what they are doing
• Methodology
• Whenever a company is mentioned in a bloggers
  post, it becomes a feature of the blogger
• Static company list used as attributes
• 1,914 company names

14
Cyclic Feature Usage
15
Blog Community Evolution
16
Blog-based Community
February 24
17
Conclusions

• Blog posts are cyclic within this community
• Post more during the week and less during the
  weekends
• Company names are mentioned in a long-tail,
  power-law way
• Few companies are mentioned often
• Most companies are mentioned rarely
• Niche communities that focus on long-tail
  companies can be discovered using IANs
• The connectivity of the blog-based IAN appears to
  follow power-law
• like explicit relationship social networks

18
Future Work

• Allow for dynamic feature extraction
• Allow features to naturally decay with time
• Putnams puzzle
• Adapt Social Capital measures to allow for
  uncorrelated bonding and bridging