Algorithm Complex Network Community Structure Mining

Complex network is an abstract representation of complex system, the study of complexnetworks is helpful for understanding complex systems. Community structure is one of thecharacteristics of complex networks topology，it is a common phenomenon that complex networkhas community structure in the real world. Mining community structure hidden in complexnetworks, contribute to a better understanding of the nature and function of complex networks,and provide a theoretical basis for the practical application of information push, personalizedservice. The value of mining community structures of complex networks both in scientific andpractical has made it become a hot field of current scientific research.This paper firstly introduces some related theoretical knowledge of mining communitystructure, then expounds some classical algorithms of community structure mining, advantagesand disadvantages of these algorithms and the application scope and summarized. Mostalgorithms have deficiencies: some high time complexity, and some results of low accuracy. Howto design a fast and reasonable network mining algorithm is still a challenging task.The main job of this paper has:(1) Based on the fusion of local node information and global module clustering idea, a newmethod named BNS algorithm is proposed to solve the problem of the community detection incomplex networks, which is based on node similarity. Firstly, the node similarity functions, theoptimal neighbor nodes according to the node similarity, merge nodes to form a small community;and then from the thought of CNM algorithm, cohesion of Societies for modularity optimization,complete community mining. To test the algorithm using the reference network, the algorithm isfeasible, computational efficiency and accuracy is improved.(2) Based on fully considering the local information, a new method named BDCN algorithmis proposed to detect community structure of complex networks， which is based on nodecentrality. Firstly, the algorithm selects the node center network degree maximum node as initialnodes; then calculate the similarity between the known node and its neighbor nodes, nodeselection of maximum similarity to known associations, to judge whether the node is added intothe community to use local modularity, achieving the purpose of community mining. Simulationexperiments using benchmark network, the results prove the feasibility and effectiveness of thealgorithm.