Mining Heterogeneous Network

Transcription

1 Mining Heterogeneous Network Clustering and Ranking Jiatu Shi Data Mining April 13, 2015 Jiatu Shi (Data Mining Mining Heterogeneous Network April 13, / 23

2 Outline 1 Introduction Why Heterogeneous? 2 Bipartite Graph: Conferences and Authors Model Methods Ranking Methods Clustering Methods Clustering Result 3 NetCLus in Star Network Schema Model Refinement Results Refinement Jiatu Shi (Data Mining Lab@UESTC) Mining Heterogeneous Network April 13, / 23

3 Outline Introduction 1 Introduction Why Heterogeneous? 2 Bipartite Graph: Conferences and Authors Model Methods Ranking Methods Clustering Methods Clustering Result 3 NetCLus in Star Network Schema Model Refinement Results Refinement Jiatu Shi (Data Mining Lab@UESTC) Mining Heterogeneous Network April 13, / 23

4 A Little Story Introduction Why Heterogeneous? One day, Prof.Han asked his students to find out his rank in computer scientists from the DBLP database( 1.8M papers, 0.7M authors, 10K venues, 70K terms). As he feel honor with his outstanding contribution in data mining. Jiatu Shi (Data Mining Mining Heterogeneous Network April 13, / 23

5 A Little Story Introduction Why Heterogeneous? But, Prof.Han felt frustrated when he saw the following results. Jiatu Shi (Data Mining Mining Heterogeneous Network April 13, / 23

6 A Little Story Introduction Why Heterogeneous? So Prof.Han had a talk with Dr.Sun. Jiatu Shi (Data Mining Lab@UESTC) Mining Heterogeneous Network April 13, / 23

7 A Little Story Introduction Why Heterogeneous? Prof.Han felt happy when he saw the result found by Dr.Sun. Jiatu Shi (Data Mining Mining Heterogeneous Network April 13, / 23

8 Introduction Basic Concepts of Network Why Heterogeneous? Jiatu Shi (Data Mining Mining Heterogeneous Network April 13, / 23

9 Introduction Basic Concepts of Network Why Heterogeneous? Jiatu Shi (Data Mining Mining Heterogeneous Network April 13, / 23

10 Outline Bipartite Graph: Conferences and Authors 1 Introduction Why Heterogeneous? 2 Bipartite Graph: Conferences and Authors Model Methods Ranking Methods Clustering Methods Clustering Result 3 NetCLus in Star Network Schema Model Refinement Results Refinement Jiatu Shi (Data Mining Lab@UESTC) Mining Heterogeneous Network April 13, / 23

11 Bipartite Graph: Conferences and Authors Model Methods Weight Matrix and Conditional Rank Jiatu Shi (Data Mining Mining Heterogeneous Network April 13, / 23

12 Bipartite Graph: Conferences and Authors Model Methods Jiatu Shi (Data Mining Mining Heterogeneous Network April 13, / 23

13 Ranking Methods Bipartite Graph: Conferences and Authors Ranking Methods Jiatu Shi (Data Mining Mining Heterogeneous Network April 13, / 23

14 Ranking Methods Bipartite Graph: Conferences and Authors Ranking Methods Jiatu Shi (Data Mining Mining Heterogeneous Network April 13, / 23

15 Mixture Model Bipartite Graph: Conferences and Authors Clustering Methods Jiatu Shi (Data Mining Mining Heterogeneous Network April 13, / 23

16 Bipartite Graph: Conferences and Authors Algorithm Framework Clustering Methods Jiatu Shi (Data Mining Mining Heterogeneous Network April 13, / 23

17 Bipartite Graph: Conferences and Authors Visualization of Clustering Clustering Methods Jiatu Shi (Data Mining Mining Heterogeneous Network April 13, / 23

18 Clustering Result Bipartite Graph: Conferences and Authors Clustering Result Jiatu Shi (Data Mining Mining Heterogeneous Network April 13, / 23

19 Bipartite Graph: Conferences and Authors Complexity Analysis Clustering Result Jiatu Shi (Data Mining Mining Heterogeneous Network April 13, / 23

20 Outline NetCLus in Star Network Schema 1 Introduction Why Heterogeneous? 2 Bipartite Graph: Conferences and Authors Model Methods Ranking Methods Clustering Methods Clustering Result 3 NetCLus in Star Network Schema Model Refinement Results Refinement Jiatu Shi (Data Mining Lab@UESTC) Mining Heterogeneous Network April 13, / 23

21 Star Network NetCLus in Star Network Schema Model Refinement Jiatu Shi (Data Mining Mining Heterogeneous Network April 13, / 23

22 NetCLus in Star Network Schema Refinement of Methods Model Refinement Jiatu Shi (Data Mining Mining Heterogeneous Network April 13, / 23

23 NetCLus in Star Network Schema Refinement of Results Results Refinement Jiatu Shi (Data Mining Mining Heterogeneous Network April 13, / 23