Research Of Evolution Social Network Analysis Method Based On Probabilistic Factor Model

Evolution Social Network is a social network with dynamically updating. The development of communications and the development of Internet interaction platform technologies have made it very easy to exchange information. Usually in these platforms, the structure of social network is highly dynamic, and their inner structural feature is richer than traditional network. Research about Evolution Social Network is still in early stages, but a great deal of achievements about static network and graph theory can be incorporate into it. Based on a lot of related analytical model, this paper analyzed evolution social network in theory and studied it in experiment.A new framework about evolution social network analysis, a new method about social network change detection and a new method for social network communities’ partition have been proposed here. All the work in this paper can be s summarized below:1. Anew framework about evolution social network analysis has been proposed. Based on factor model, it optimizes the connection eigenvector, rebuilds the connection matrix and calculates the evolution index.2. A new definition from probabilistic perspective about evolution social network has been proposed. Based on the definition, this paper proposed a probabilistic factor model for calculating the evolution index.3. A new method based on probabilistic methodology for social network community partition has been proposed. Followed with heuristic algorithms in correlation clustering, this paper divided the communities and evaluated the result.In the foundation of above theoretical analysis, Experiments in random datasets and real-world datasets has been put into practice. Experiments show that, Event-Detect algorithm Prob-Event-Detect algorithm effectively detected structural transformation in evolution social network. In comparing experiments, Prob-Cluster-Detect algorithm showed robustness and reasonable result in community detecting.Lastly, the research is summarized and the future extensions of the relevant study are presented. For example:time-space model in evolution social network, expanding connection model in evolution social network and evolution social network with relationship.