Research On Signed Social Network Local Community Detection Algorithm

With the rapid development of Internet, various network application platforms have gradually infiltrated into people’s daily life and study. Signed network as a branch of complex network system, the scale of data is more and more large, how to mine valuable information quickly and effectively becomes important. Community detection is the basis of researches of complex networks, the researches and applications on personalized recommendation and analysis of characteristics user are dependent on it.For signed network local community detection, the problems to be solved are how to improve the accuracy and stability of the algorithm. Based on the current status on local community detection, according to the existing problems, the paper researches as follows:Firstly, in the algorithm processed using two phases, for the problem that the community recognition accuracy is not high which is caused by losing some of negative edge information, introduces the concepts of contribution degree and interference degree, combines the link signs and link density, considers the participation degree synthetically when the node participates a community, adjusts the visited order of the nodes, proposes the community detection algorithm TFA based on two-phase fusion.Secondly, for the problem that the algorithm stability is not high which is caused by choosing the start node randomly, combines the nodes and neighbor’s attribute information and structural balance theory, uses the node influence and clustering coefficient, puts forward a new strategy of selecting the start node; based on this, introduces node similarity, considers the relationship between nodes, proposes the community detection algorithm NSA based on node similarity.Finally, the TFA algorithm is verified by comparing the experiment, the tests show that the algorithm shows good performance in the community recognition accuracy and running time, especially when handling the larger networks. Besides, through experiments, The paper also proves the accuracy of NSA.