a framework for finding communities in dynamic social networks david kempe university of southern...
TRANSCRIPT
![Page 1: A Framework for Finding Communities in Dynamic Social Networks David Kempe University of Southern California Chayant Tantipathananandh, Tanya Berger-Wolf](https://reader036.vdocuments.net/reader036/viewer/2022070411/56649f535503460f94c77e98/html5/thumbnails/1.jpg)
A Framework for Finding Communities in Dynamic Social
Networks
David KempeUniversity of Southern California
Chayant Tantipathananandh, Tanya Berger-WolfUniversity of Illinois at Chicago
![Page 2: A Framework for Finding Communities in Dynamic Social Networks David Kempe University of Southern California Chayant Tantipathananandh, Tanya Berger-Wolf](https://reader036.vdocuments.net/reader036/viewer/2022070411/56649f535503460f94c77e98/html5/thumbnails/2.jpg)
Social Networks
![Page 3: A Framework for Finding Communities in Dynamic Social Networks David Kempe University of Southern California Chayant Tantipathananandh, Tanya Berger-Wolf](https://reader036.vdocuments.net/reader036/viewer/2022070411/56649f535503460f94c77e98/html5/thumbnails/3.jpg)
History of Interactions
t=1
History of interactions
t=1t=1
t=2t=2
t=3t=3
t=4t=4
t=5t=51122 334455
55 44 1122 33
55 22 33 44 11
55 22 33 44
55 22 44 11
12 3
45
Assume discrete time and interactions in form of complete subgraphs.
Aggregated Aggregated networknetwork
5
4
23
1
2
3 2
1
11
![Page 4: A Framework for Finding Communities in Dynamic Social Networks David Kempe University of Southern California Chayant Tantipathananandh, Tanya Berger-Wolf](https://reader036.vdocuments.net/reader036/viewer/2022070411/56649f535503460f94c77e98/html5/thumbnails/4.jpg)
Community Identification
• Centrality and betweenness [Girvan & Newman ‘01]
• Correlation clustering [Basal et al. ‘02]
• Overlapping cliques [Palla et al. ’05]
What is community?
“Cohesive subgroups are subsets of actors among whom there are relatively strong, direct, intense, frequent, or positive ties.” [Wasserman & Faust ‘97]
Notions of communities:
Static
Dynamic
• Metagroups [Berger-Wolf & Saia ’06]
![Page 5: A Framework for Finding Communities in Dynamic Social Networks David Kempe University of Southern California Chayant Tantipathananandh, Tanya Berger-Wolf](https://reader036.vdocuments.net/reader036/viewer/2022070411/56649f535503460f94c77e98/html5/thumbnails/5.jpg)
The Question: What is dynamic community?
•A dynamic community is a subset of individuals that stick together over time.
•NOTE: Communities ≠ Groups
5 4 32 1
5
4
5
4
1
4
12 3 4
5 2
2 3
5 2 3 1
t=1
t=2
t=3
t=4
t=5
![Page 6: A Framework for Finding Communities in Dynamic Social Networks David Kempe University of Southern California Chayant Tantipathananandh, Tanya Berger-Wolf](https://reader036.vdocuments.net/reader036/viewer/2022070411/56649f535503460f94c77e98/html5/thumbnails/6.jpg)
Approach: Graph Model
5
5
5
5
5
1 2 3 4
1 2 3 4
1 2 3 4
1 2 3 4
1 2 3 4
t=1
t=2
t=3
t=4
t=5 1122 334455
55 44 1122 33
55 22 33 44 11
55 22 33 44
55 22 44 11
![Page 7: A Framework for Finding Communities in Dynamic Social Networks David Kempe University of Southern California Chayant Tantipathananandh, Tanya Berger-Wolf](https://reader036.vdocuments.net/reader036/viewer/2022070411/56649f535503460f94c77e98/html5/thumbnails/7.jpg)
Approach: Assumptions
• Individuals and groups represent exactly one community at a time.
• Concurrent groups represent distinct communities.
Desired
Required
•Conservatism: community affiliation changes are rare.
•Group Loyalty: individuals observed in a group belong to the same community.
•Parsimony: few affiliations overall for each individual.
![Page 8: A Framework for Finding Communities in Dynamic Social Networks David Kempe University of Southern California Chayant Tantipathananandh, Tanya Berger-Wolf](https://reader036.vdocuments.net/reader036/viewer/2022070411/56649f535503460f94c77e98/html5/thumbnails/8.jpg)
Approach: Color = Community
Valid coloring: distinct color of groups in each time step
![Page 9: A Framework for Finding Communities in Dynamic Social Networks David Kempe University of Southern California Chayant Tantipathananandh, Tanya Berger-Wolf](https://reader036.vdocuments.net/reader036/viewer/2022070411/56649f535503460f94c77e98/html5/thumbnails/9.jpg)
Approach: Assumptions
• Individuals and groups represent exactly one community at a time.
• Concurrent groups represent distinct communities.
Desired
Required
•Conservatism: community affiliation changes are rare.
•Group Loyalty: individuals observed in a group belong to the same community.
•Parsimony: few affiliations overall for each individual.
![Page 10: A Framework for Finding Communities in Dynamic Social Networks David Kempe University of Southern California Chayant Tantipathananandh, Tanya Berger-Wolf](https://reader036.vdocuments.net/reader036/viewer/2022070411/56649f535503460f94c77e98/html5/thumbnails/10.jpg)
Costs
•Conservatism: switching cost (α)
•Group loyalty:-Being absent (β1) -Being different (β2)
•Parsimony: number of colors (γ)
![Page 11: A Framework for Finding Communities in Dynamic Social Networks David Kempe University of Southern California Chayant Tantipathananandh, Tanya Berger-Wolf](https://reader036.vdocuments.net/reader036/viewer/2022070411/56649f535503460f94c77e98/html5/thumbnails/11.jpg)
Approach: Assumptions
• Individuals and groups represent exactly one community at a time.
• Concurrent groups represent distinct communities.
Desired
Required
•Conservatism: community affiliation changes are rare.
•Group Loyalty: individuals observed in a group belong to the same community.
•Parsimony: few affiliations overall for each individual.
![Page 12: A Framework for Finding Communities in Dynamic Social Networks David Kempe University of Southern California Chayant Tantipathananandh, Tanya Berger-Wolf](https://reader036.vdocuments.net/reader036/viewer/2022070411/56649f535503460f94c77e98/html5/thumbnails/12.jpg)
Costs
•Conservatism: switching cost (α)
•Group loyalty:-Being absent (β1) -Being different (β2)
•Parsimony: number of colors (γ)
![Page 13: A Framework for Finding Communities in Dynamic Social Networks David Kempe University of Southern California Chayant Tantipathananandh, Tanya Berger-Wolf](https://reader036.vdocuments.net/reader036/viewer/2022070411/56649f535503460f94c77e98/html5/thumbnails/13.jpg)
Approach: Assumptions
• Individuals and groups represent exactly one community at a time.
• Concurrent groups represent distinct communities.
Desired
Required
•Conservatism: community affiliation changes are rare.
•Group Loyalty: individuals observed in a group belong o the same community.
•Parsimony: few affiliations overall for each individual.
![Page 14: A Framework for Finding Communities in Dynamic Social Networks David Kempe University of Southern California Chayant Tantipathananandh, Tanya Berger-Wolf](https://reader036.vdocuments.net/reader036/viewer/2022070411/56649f535503460f94c77e98/html5/thumbnails/14.jpg)
Costs
•Conservatism: switching cost (α)
•Group loyalty:-Being absent (β1) -Being different (β2)
•Parsimony: number of colors (γ)
![Page 15: A Framework for Finding Communities in Dynamic Social Networks David Kempe University of Southern California Chayant Tantipathananandh, Tanya Berger-Wolf](https://reader036.vdocuments.net/reader036/viewer/2022070411/56649f535503460f94c77e98/html5/thumbnails/15.jpg)
Problem Definition
•Minimum Community Interpretation For a given cost setting, (α,β1,β2,γ), find vertex coloring that minimizes total cost.
• Color of group vertices = Community structure
• Color of individual vertices = Affiliation sequences
• Problem is NP-Complete and APX-Hard
![Page 16: A Framework for Finding Communities in Dynamic Social Networks David Kempe University of Southern California Chayant Tantipathananandh, Tanya Berger-Wolf](https://reader036.vdocuments.net/reader036/viewer/2022070411/56649f535503460f94c77e98/html5/thumbnails/16.jpg)
Model Validation and Algorithms
• Model validation: exhaustive search for an exact minimum-cost coloring.
• Heuristic algorithms evaluation: compare heuristic results to OPT.
• Validation on data sets with known communities from simulation and social research
- Southern Women data set (benchmark)
![Page 17: A Framework for Finding Communities in Dynamic Social Networks David Kempe University of Southern California Chayant Tantipathananandh, Tanya Berger-Wolf](https://reader036.vdocuments.net/reader036/viewer/2022070411/56649f535503460f94c77e98/html5/thumbnails/17.jpg)
Southern Women Data Set
by Davis, Gardner, and Gardner, 1941
Photograph by Ben Shaln, Natchez, MS, October; 1935 Aggregated network
Event participation
![Page 18: A Framework for Finding Communities in Dynamic Social Networks David Kempe University of Southern California Chayant Tantipathananandh, Tanya Berger-Wolf](https://reader036.vdocuments.net/reader036/viewer/2022070411/56649f535503460f94c77e98/html5/thumbnails/18.jpg)
Ethnographyby Davis, Gardner, and Gardner, 1941
Core (1-4)
Periphery (5-7)
Core (13-15)
Periphery (11-12)
![Page 19: A Framework for Finding Communities in Dynamic Social Networks David Kempe University of Southern California Chayant Tantipathananandh, Tanya Berger-Wolf](https://reader036.vdocuments.net/reader036/viewer/2022070411/56649f535503460f94c77e98/html5/thumbnails/19.jpg)
An Optimal Coloring: (α,β1,β2,γ)=(1,1,3,1)
Cor
eP
erip
hery
Pe
riph
ery
Cor
e
![Page 20: A Framework for Finding Communities in Dynamic Social Networks David Kempe University of Southern California Chayant Tantipathananandh, Tanya Berger-Wolf](https://reader036.vdocuments.net/reader036/viewer/2022070411/56649f535503460f94c77e98/html5/thumbnails/20.jpg)
An Optimal Coloring: (α,β1,β2,γ)=(1,1,1,1)
Cor
eP
erip
he
ry
Cor
e
![Page 21: A Framework for Finding Communities in Dynamic Social Networks David Kempe University of Southern California Chayant Tantipathananandh, Tanya Berger-Wolf](https://reader036.vdocuments.net/reader036/viewer/2022070411/56649f535503460f94c77e98/html5/thumbnails/21.jpg)
Conclusions
• An optimization-based framework for finding communities in dynamic social networks.
• Finding an optimal solution is NP-Complete and APX-Hard.
• Model evaluation by exhaustive search.
• Heuristic algorithms for larger data sets. Heuristic results comparable to optimal.
![Page 22: A Framework for Finding Communities in Dynamic Social Networks David Kempe University of Southern California Chayant Tantipathananandh, Tanya Berger-Wolf](https://reader036.vdocuments.net/reader036/viewer/2022070411/56649f535503460f94c77e98/html5/thumbnails/22.jpg)
Thank You
Poster #6 this evening
![Page 23: A Framework for Finding Communities in Dynamic Social Networks David Kempe University of Southern California Chayant Tantipathananandh, Tanya Berger-Wolf](https://reader036.vdocuments.net/reader036/viewer/2022070411/56649f535503460f94c77e98/html5/thumbnails/23.jpg)
Dan RubensteinPrinceton
Siva Sundaresan
Ilya Fischoff
Simon LevinPrinceton
David KempeUSC
Jared SaiaUNM
MuthuGoogleHabib
a
Mayank Lahiri
Computational PopulationBiology Lab
UIC
compbio.cs.uic.edu
TanyaBerger-Wolf
ChayantTantipathananand
h
Poster#6this evening