Community Detection In Online Personal Social Network Visual Analysis System

More and more people build their own personal homepage on social networkingsites to communicate with others online, forming huge social networks.A wealth ofonline social networking resources is significant for social network research.How tohelp users to better observe and understand their own online social network, whichhide valuable information, is an interesting and urgent problem that should be solved.On the basis of a comprehensive exposition of the common social networkanalysis methods, this thesis focuses on community detection problem.In the processof community detection, the traditional hierarchical clustering algorithm need tocalculate the similarity between all pairs of vertices. Aimed at this shortcoming, theconcept of "Ego" role in the Ego Networks is introduced, and the detecting process ofthe "Ego" role is added to the process of the traditional hierarchical clusteringalgorithm. Then the similarity between all other vertices and the "Ego" vertice iscalculated, to improve community detection efficiency. At last, the effectiveness ofthe new algorithm process is verified in different types of real networks.On the basis of a more comprehensive understanding of social network analysistheory and renren.com, the thesis proposes the concept of online personal socialnetwork visual analysis system, called Vizrenren. The detail of the system architectureof Vizrenren has been described, and relevant tools are used to simulate the functionof each module. Finally, based on Renren API and open source software Vizster,Vizrenren is implemented. The system test results show that the tool can help ordinaryonline social network users to observe and discover their own online social network ina interesting way.