The Research Of Community Partitioning Algorithm Based On Spectral Bisection Method

Many actual systems in nature are able to be reflected by a variety of network. During the process of analysis and research community structure, people found that the community structure is the common property of many networks. Study on community structure of networks has very important theoretical significance and broad application prospects to analyze the topology of complex network, understand its features, find hidden patterns and predict network behavior.At present people have proposed many algorithms to solve the problems on how to find the community structure of complex networks. In this thesis, we propose two new node similarity matrixes, and then use the spectral bisection method and FCM methods to make division of community structure of complex networks. Research done as follows:This thesis presents two new community partition algorithm, respectively using the thought of the neighbor nodes connection degree and the shortest path, combining SNN similarity matrix with spectral bisection method, and then to divide complex network community. We test the proposed algorithm through computer-generated network and actual network, and the experimental results show that the algorithm can be good found the network community when network community structure is not obvious.