Visualizing social network concepts

bleu.tw@gmail.com

Transcript

1. Visualizing social network concepts Decision Support Systems 49 (2010) 151–161 49941131, Jia-Huei Ren bleu.tw@gmail.com
2. Journal • Decision Support Systems 49 (2010) 151–161 • Author : Bin Zhu , Stephanie Watts, Hsinchun Chen • Impact Factor : 2.036 • RANK : Category Name Total Journal Journal Rank Quartile COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE 121 27 Q1 COMPUTER SCIENCE, INFORMATION SYSTEMS 135 20 Q1 OPERATIONS RESEARCH & MANAGEMENT SCIENCE 79 9 Q1
3. You may think social network is…
4. But actually…
5. That’s not social network
6. Social network is…
7. Visualizing?
8. Motivation • Social network concepts are invaluable for understanding the social network phenomena. • Most existing network visualization techniques provide limited support for the comprehension of network concepts.
9. Objective • To propose an approach called concept visualization to facilitate the understanding of social network concepts.
10. Methods • Web crawler • Text Mining – Natural Language Processing • Social Network Clustering – Circular SOM
11. Question • 此系統尚未使用下列何種工具/技術? – 1. Web crawler – 2. Internet of things – 3. Text mining
12. Web Crawler
13. Internet Web Crawling
14. Text Mining
15. Text Input
16. Convert into a structured format • 2014/11/24 • Nov 24,2014 • 2014.11.24 • 2014-11-24 20141124 • Wow, Cathy is so beautiful! • Wow, Cathy is so beautiful beauty!
17. Classifying, Clustering
18. Analyze
19. Natural Language Processing
20. Social Network Clustering and Visualizing
21. What is SOM? • Unsupervised Learning Network • N-dimension -> 2-dimension
22. What is Circular SOM?
23. Comparison Table Number Of Clusters (K) Performance (Lower is better) SOM K-Means EM HCA 8 59 63 62 65 16 67 71 69 74 32 78 84 84 87 64 85 89 89 92
24. Betweenness centrality
25. Highest Betweenness “Bridge” “Convenience”
26. Highest Closeness 1. Find shortest path lengths to others 2. Take average of these 3. Closeness = 1/ (Average shortest path lengths)
27. Degree Centrality Highest Degree
28. Out-Degree Highest Out-Degree “Gregarious”
29. In-Degree Highest In-Degree “Popular”
30. Structural similarity
31. Gatekeepers of subgroups
32. The Network visualization using the conventional approach
33. The interface of the NetVizer System
34. Task Design Network concepts Task Degree centrality 1. Count links of given actor 2. Compare number of link of two actors 3. Find actors with the maximum number of links Structural equivalence 4. Who knows number of friends of a given actor ? Group identification 5. If given actor belongs to a group ? Interaction between groups 6. Identify gatekeepers of a given groups Betweenness centrality 7. Which one of the two actors should be removed to disconnect more people in the network ?
35. Design/Operating principle Set Keywords Crawlering Text Mining Analysis Visualizing Existing Database
36. Conclusions  The proposed concept visualization approach that explicitly presents the network concepts of degree centrality, betweenness centrality, subgroup identification, gatekeepers, and structural equivalence.
37. Comments • Advantage – Help for understanding of social network concepts • Application – Network visualization – Social network analysis – Information categorization – Information analysis
38. Further works • Other domains • Effectiveness • Automatic
39. References • Osama Abu Abbas(2008), “Comparisons Between Data Clustering Algorithms” • Da-Yu Yuan, Muh-Chyun Tang(2000) , “Exploring Intellectual Network Structure of an Interdisciplinary Research Community: A Case Study of Taiwan’s STS Community” • Bala Deshpande(2012), “3 ways to use text mining with RapidMiner to juice up your job search” • 陳信宇(2010), “Reinforcement Learning of Robot: Integrating Genetic Programming and Neural Network” • Bin Wang(2014), “Visualizing Multivariate Network on The Surface of A Sphere”
40. Thanks