Community Discovery And Tracking Methods Based On Core Members

Social network is a combination of individuals,or organizations,as well as their connections.In large-scale social networks,some sociology features will become more obvious.In recent years,as the rapid development of the internet and social network sites,the study of large-scale social network has become hot.Our main task is to study the current algorithms of community discovery and the main problems encountered, then we propose a algorithm based on core member to detect and track the evolution of community, by conducting experiments and comparative experiments, analysis of the model is correct. Resulted in the following research areas:(1) We propose a core members based community discovery algorithm. First identify the core members of the network, and then we merge non-core member vertex to core member vertex which is most similar to it, and finally using the Average linkage algorithm to merge the results of the initial division of the community, to get the final results of the community divided. The proposed algorithm can be controlled by adjusting the threshold value. By adjusting the threshold we can balance time complexity and the quality of community discovery.(2) Analyzing the dynamic nature of online communities. First, A large community will survive long period of time; the second is a large and stable members of the community is low, most of the members existed in one or two consecutive times, only a small part of the core members have a longer life cycle.(3) We proposed a model based on the core members to track the evolution of the community. Based on the result of community discovery,we proposed a method to track the evolution of the community. The experiment shows the effectiveness of the algorithm to track the evolution of the reasons for the community.By adjusting the threshold we can balance time complexity and the quality of community discovery. Experiments show this algorithm can not only improve the efficiency but also keep high quality of community discovery. It is the effective method for processing large amount of data.