Research On Community Detection Algorithm Base On Fuzzy Clustering

With the develop of the modern network science, people find that many networks have some common properties,which the most important property called community structure. The community structure reveals the structure of complex network and be of important theoretical significance in analysis of functions in complex network and be helpful in understanding the network. Now, the community structure of network has been the focus of the network research and get extensive researched.The clustering analysis is an extraordinary important technology in pattern recognition and system modeling. It is widely used in the fields of data-mining, artificial intelligence, machine study. There are many clustering algorithms among which fuzzy c-means(FCM) gets the most concerns because of its complete theory. Based on the FCM algorithm, the paper combines with the others algorithms to improve the FCM algorithm and applies the proposed algorithm in the community detection of network. The main researches of this paper are listed as follows:1.The paper proposes an improve FCM algorithm based on data field. The improved algorithm uses the data field according the theory of fields in physics to determine the initial class centers, avoided the algorithm get into the local extremum and reduced the number of iteration. The improved algorithm proposes clusters merging to avoid the set of the number of clustering and broad the applied scopes. The simulation experiment results show that the improved algorithm could make up the defects of traditional FCM algorithm and improve the efficiency and accuracy of clustering.2.In the view of the phenomenon of a lot of overlapping community in the network. The paper devises the algorithm of overlapping community detection based on FCM algorithm. The algorithm gets center nodes based on topological potential to avoid the random of the result. The algorithm clusters the network by FCM algorithm to get the overlapping community in the network. The algorithm carry on the community merging to find the better community structure, improved the rationality of community detection. The number of found communities is closer to the real number of communities. The simulation experiment results demonstrate that the algorithm is efficient for detecting overlapping community in the network and the algorithm has better partitioning ability and lower complexity.