0
A Structural Approach to
Community-level Social
Influence Analysis
Ph.D. Viva

Václav Belák
Context and Motivation I
Our earlier study suggested communities influence each other

2 / 25
Context and Motivation II
• Network represents flow
between actors

• Actor-level social influence in
healthcare, innovati...
Research Problem and Questions
Problem: measurement, analysis, and explanation of influence
between various types of socia...
Q1: How can we model influence between communities?

5 / 25
Methodology: COIN
What
impacts

depends on

How
T

centrality

communities

communities

membership

communities

actors

...
Impact and Its Aggregates
impacts

•
•
•
•
•
•

communities

depends on

communities

Σ

Σ

row – impact of a community on...
Experiments

8 / 25
Influence Over Time
Questions:
• Which communities influenced a given community over time?
• How do we measure that by COI...
Personal Issues vs Moderators
emphasised:
strong impact
impacting forum

impact

10

●

Personal Issues
Moderators

5
PI M...
Q2: How do we detect communities acting as global authorities/hubs?

11 / 25
importance

Global Authorities:
Widespread High Importance
local
authorities

global
authorities

low

widespread
low

imp...
Moderators: Authority of
●

importance

2.0

Moderators
●

1.5
●
●
●

●
●

1.0

●

●

●

●

●

●
●

0.5

●

●

●
●
● ● ●
●...
Global Hubs:
Widespread High Dependence
hubs

low

widespread low

dependence

driven

dependence entropy
14 / 25
After Hours: Hub of

dependence

●

10

After Hours

5
●
●
●
●

●

●

●

●

●

●
●
● ●●
●●
●
●
●
●
●
●
● ● ● ● ●● ● ● ● ●
...
Core: Hub of

dependence

 COIN integrated to SAP PULSAR

SAP Business
One: Core

dependence entropy
16 / 25
Cross-Community Dynamics in Science
Questions
• How can we measure and explain
influence between scientific
communities?
•...
COIN for Scientific Communities
• citations as a proxy of impact and information flow
citation

information flow

Aggregat...
Exporters and Isolated AI Communities
Hypothesis
• importance indicates exporters
• independence and importance indicat...
Q3: Can we exploit the model to maximise information diffusion?

20 / 25
Influence and Information Diffusion
high in-degree

Cross-community diffusion maximisation problem:
Actor-level diffusion ...
Information Diffusion Experiments
• Hypothesis: product of importance and entropy identifies seed
communities that induce ...
Selection

user activation fraction (a)

COIN Optimises Information Diffusion
0.05
●

0.04

●
●

0.03
0.02
0.01

●
●

1

u...
Summary and Future Work
•
•
•
•
•

COIN: computational model for community influence
Communities influencing a particular ...
Contributions
• proposes a solution to the problem of measurement, analysis, and
explanation of influence between communit...
Personal Issues and Moderators

membership

indegree

1.00

●

●

0.75

ld
12

●

30

●
●

8

●

20

0.50
0.25

10

0.00

...
CBR community: isolated
CBR
in−, out−flow

1.6
1.2
●
●

●
●

out−flow
in−flow
introspection

●
●

out−flow
in−flow
introsp...
CBR: isolated and shrinking
• decreasing size
• rigid member-base

rising impact factor
driven by self-citations
●
●

●
●
...
Greedy Strategy

29
Group In-Degree

GI = # links from outside

30
•
•
•
•

COIN extended to capture topics
Based on tensor algebra
Better interpretability and sensitivity
Consistent with p...
Rise of Hubs and Authorities in Boards

32
Exporters and Introspective Communities

33
Upcoming SlideShare
Loading in...5
×

Vaclav Belak PhD Viva

157

Published on

Published in: Education
0 Comments
1 Like
Statistics
Notes
  • Be the first to comment

No Downloads
Views
Total Views
157
On Slideshare
0
From Embeds
0
Number of Embeds
1
Actions
Shares
0
Downloads
5
Comments
0
Likes
1
Embeds 0
No embeds

No notes for slide
  • I’m going to present how to use structure of social interactions to quantify and explain influence between communities.
  • topics flow between communitiescommunities may have a position suggesting an important role as a “bridging” community“Wouldn’t you want to know whether the community you regularly engage with as a researcher is gaining or loosing influence? My research provides answers to such questions.”
  • network represent flow, e.g. frequent information exchangein-degree: frequently responded actor (e.g. cited) is influentialreply as activity stimulationreply as information flowhigh in-degree: control over flow
  • HITS cannot be used to address these questions because it is global measure 1 node vs rest
  • methodological core of our model: COmmunityINfluenceHypothesisof cross-comm impactInfluence measured by impactMembership – distribution of engagement, core vs restCentrality ~ position: control over flow of resource, high/low Cin-degree: tendency to stimulateInfluence ~ stimulation of responses (citations, replies, etc.) by the core members: high/low JDependence ~ community’s activity is driven by core members of other communities
  • independence used to threshold strong impact – community influences activity more than the community itself
  • Boards: 10 yearsSAP: 8 years
  • HITS is a node-level measure and cannot be applied
  • 19 years of data
  • middle period: 1997-2002COLT – strong exporter, Conference on Learning TheoryIJCAI – exports, but consists of core members of other communitiesCBR – isolated, may lead to decline: hard to get external resources like funding or attract new memberswe have much more supportive evidence that CBR declined: size or citation impact
  • Actor-level: Application in public health, marketing, innovation managementCommunity-level: online fora, conferences, any mass-medium; recently gained more attentionSimulation used to simulate the spread
  • Part of the resultsWeek 497, uss=1
  • JELIA - European Conference on Logics in Artificial Intelligence
  • size as a cardinality of the set of the membersdecrease in # papersdecrease in Google Trends since 2005
  • ----- Meeting Notes (17/02/2014 19:16) -----remove or fix
  • 13 strong impacts V-TFL Admin -> V-TFL Discussion
  • Transcript of "Vaclav Belak PhD Viva"

    1. 1. A Structural Approach to Community-level Social Influence Analysis Ph.D. Viva Václav Belák
    2. 2. Context and Motivation I Our earlier study suggested communities influence each other 2 / 25
    3. 3. Context and Motivation II • Network represents flow between actors • Actor-level social influence in healthcare, innovations, marketing, etc. high in-degree • Actors embedded in communities • No suitable model of community-level influence 3 / 25
    4. 4. Research Problem and Questions Problem: measurement, analysis, and explanation of influence between various types of social communities Questions 1. How can we model influence between communities? 2. How do we detect communities acting as global authorities/hubs? 1. Can we exploit the model to maximise information diffusion? 4 / 25
    5. 5. Q1: How can we model influence between communities? 5 / 25
    6. 6. Methodology: COIN What impacts depends on How T centrality communities communities membership communities actors actors communities impact 6 / 25
    7. 7. Impact and Its Aggregates impacts • • • • • • communities depends on communities Σ Σ row – impact of a community on others column – impact of others on a community diagonal – independence importance = total impact of a community on others dependence = total impact of others on a community importance/dependence heterogeneity measured by entropy 7 / 25
    8. 8. Experiments 8 / 25
    9. 9. Influence Over Time Questions: • Which communities influenced a given community over time? • How do we measure that by COIN? Hypothesis • Frequent impact higher than independence indicates influence Experiments • segment data by time window • find impact higher than independence of influenced community Discussion fora data • links represent replies • forum as a proxy of community 9 / 25
    10. 10. Personal Issues vs Moderators emphasised: strong impact impacting forum impact 10 ● Personal Issues Moderators 5 PI Mods 0 200 300 400 time  Personal Issues influenced first by Moderators  Later by a specific moderating community, PI Mods 10 / 25
    11. 11. Q2: How do we detect communities acting as global authorities/hubs? 11 / 25
    12. 12. importance Global Authorities: Widespread High Importance local authorities global authorities low widespread low importance entropy 12 / 25
    13. 13. Moderators: Authority of ● importance 2.0 Moderators ● 1.5 ● ● ● ● ● 1.0 ● ● ● ● ● ● ● 0.5 ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ●● ● ● ● ● ● ● ● ●● ● ● ●● ● ● ● ● ● ● ● ● ● ●● ● ● ●● ●● ●● ● ● ● ● ● ● ● ● ●● ● ●● ● ●●● ● ● ● ● ●●● ● ●●● ● ● ● ● ●● ● ● ●● ● ●● ● ● ●● ● ● ● ●● ●● ● ● ● ● ● ●●● ● ●● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● 0.55 0.60 ● ● ● ● ● ● 0.65 0.70 importance entropy 13 / 25
    14. 14. Global Hubs: Widespread High Dependence hubs low widespread low dependence driven dependence entropy 14 / 25
    15. 15. After Hours: Hub of dependence ● 10 After Hours 5 ● ● ● ● ● ● ● ● ● ● ● ● ●● ●● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ●● ●● ●● ●● ● ● ●● ● ● ● ● ● ● ●● ●● ●●● ●● ●●● ● ● ●● ● ● ● ●●● ● ● ●● ● ● ●● ● ●● ●● ● ● ● ● ● ●●●●● ●● ●●●●●● ●● ● ● ●●●●● ●●●●●● ●●●●●●●●●●● ●●●●●●●●● ● ● ● ● ● ● ●● ●● ●● ●●●●● ● ●● ● ●● ● ● 0 ● ● 0.4 0.5 0.6 ● ● 0.7 0.8 0.9 dependence entropy 15 / 25
    16. 16. Core: Hub of dependence  COIN integrated to SAP PULSAR SAP Business One: Core dependence entropy 16 / 25
    17. 17. Cross-Community Dynamics in Science Questions • How can we measure and explain influence between scientific communities? • How does the influence relate to community’s performance? • How do we adapt COIN? Data • Scientists linked by citations • AI communities defined as conferences 17 / 25
    18. 18. COIN for Scientific Communities • citations as a proxy of impact and information flow citation information flow Aggregate Measures • importance: how much information flows out of the community • independence: how introspective the community is 18 / 25
    19. 19. Exporters and Isolated AI Communities Hypothesis • importance indicates exporters • independence and importance indicates isolated islands ● CBR ● ● independence 0.75 islands COLT exporters 0.50 ● 0.25 mainstream ● ● ● 0.00 ● ● ● ● ● ● ● ● ● ●●● ● ● ● ● ●● ● ● 0.0 ● ● ● ● ● ● ● loose exporters ● ● ● IJCAI 0.5 1.0 importance 1.5 19 / 25
    20. 20. Q3: Can we exploit the model to maximise information diffusion? 20 / 25
    21. 21. Influence and Information Diffusion high in-degree Cross-community diffusion maximisation problem: Actor-level diffusion maximisation problem: Which communities to target? Which actors to target? 21 / 25
    22. 22. Information Diffusion Experiments • Hypothesis: product of importance and entropy identifies seed communities that induce high overall adoption • Overall adoption estimated by a diffusion model on • Four targeting strategies: 1. 2. 3. 4. Impact Focus (IF) – COIN Greedy (GR) Group In-degree (GI) Random (RA) IF = importance × entropy • Selection vs Prediction 22 / 25
    23. 23. Selection user activation fraction (a) COIN Optimises Information Diffusion 0.05 ● 0.04 ● ● 0.03 0.02 0.01 ● ● 1 user activation fraction (a) Greedy overfits Prediction strategy ● IF GI GR RA 2 3 strategy# seed communities (q) 4 0.05 Impact 5 Focus is more robust ● 0.04 strategy ● IF GI GR RA ● 0.03 ● 0.02 ● 0.01 ● 0.00 1 2 3 4 5 # seed communities (q) 23 / 25
    24. 24. Summary and Future Work • • • • • COIN: computational model for community influence Communities influencing a particular community Roles of communities: authorities vs hubs Isolated communities loosing influence Seed communities for information diffusion • General (3 systems) and extensible • Tensor-based extension of COIN captures topics Future Work  May be applicable to e.g. email networks  Impact Focus may be improved by discounting overlap  Sentiment-informed community influence 24 / 25
    25. 25. Contributions • proposes a solution to the problem of measurement, analysis, and explanation of influence between communities • purely structural approach • extended to capture topics • empirical analysis of 3 systems – common/different phenomena • first approach to novel problem of cross-community information diffusion Dissemination • 1 journal, 3 conference, and 1 workshop papers • best poster at NUIG research day 2013 • complete results, software, data, thesis, etc. at: http://belak.net/doc/2014/thesis.html 25 / 25
    26. 26. Personal Issues and Moderators membership indegree 1.00 ● ● 0.75 ld 12 ● 30 ● ● 8 ● 20 0.50 0.25 10 0.00 ● ● ● ● ● ● ● 4 0 PI PIM ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● group PI PIM ● 0 PI PIM PI PIM 26
    27. 27. CBR community: isolated CBR in−, out−flow 1.6 1.2 ● ● ● ● out−flow in−flow introspection ● ● out−flow in−flow introspection ● ● 0.8 ● ● ● ● ● ● ● ● ● ● 0.4 ● ● ● ● ● ● 1996 1997 1998 ● ● ● ● 1999 2000 2001 2002 ● ● 2003 2004 2005 2006 2007 2008 year JELIA in−, out−flow 3 ● ● 2 ● ● 1 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 0 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 year 27
    28. 28. CBR: isolated and shrinking • decreasing size • rigid member-base rising impact factor driven by self-citations ● ● ● ● group in−degree 160 ● ● size ● ● 140 ● ● ● ● ● ● ● ● 120 ● ● ● ● ● ● ● ● ● ● 120 ● ● ● ● ● ● ● ● 80 ● ● ● ● 40 ● ● ● ● ● ● ● ● 1.00 ● ● ● ● impact factor ● ● ● ● ● ● 0.75 ● ● ● ● ● ● 0.50 ● ● ● ● 0.25 ● ● ● ● ● ● 0.00 ● ● ● ● ● ● 1996 1998 2000 2002 2004 2006 2008 1996 1998 2000 2002 2004 2006 2008 1996 1998 2000 2002 2004 2006 2008 year year year  CBR was unable to attract new members and decayed  Cannot be revealed by introspective analysis 28
    29. 29. Greedy Strategy 29
    30. 30. Group In-Degree GI = # links from outside 30
    31. 31. • • • • COIN extended to capture topics Based on tensor algebra Better interpretability and sensitivity Consistent with purely structural COIN actors Topical Dimensions of Influence communities • Example: V-TFL Admin vs V-TFL Discussion 31
    32. 32. Rise of Hubs and Authorities in Boards 32
    33. 33. Exporters and Introspective Communities 33
    1. A particular slide catching your eye?

      Clipping is a handy way to collect important slides you want to go back to later.

    ×