An Algorithm For Detecting Community Structure Via Node Similarity

The content is no longer narrowly sociological research in today's social networks.It turns into a set of cutting-edge scientific value and great commercial potential in one of the hot research topic and attracting more and more attention of researchers in various fields.With the development of the times, the data in the Internet are also rapidly increasing. Big Data era of network has become extremely complex. Community structure is a complex network of the most important characteristics which consists of a series of nodes and edges.Its community internal node connections are tight, community connections between nodes are sparse. The research focused on how to pinpoint the hidden structure in a complex net work.The existing community discovery algorithms exist many problems. It is hard to overcome the problem that have time efficiency and community recognition accuracy at the same time. Because pursuit of division accuracy rate decrease will lead to the identification efficiency. The optimization algorithm running speed but lead to lack of precision on the algorithm. In addition, Some algorithms also rely on prior knowledge of the properties and characteristics of the network to be divided by the number of communities and not to identify the structure of spontaneous communities through structural characteristics of the network but relys on artificial setting parameters. Therefore, we propose a community based on the node similarity discovery algorithm. Community moderate maximum node as the starting node,and by similarity measure function to find the node with the highest similarity. This forms an initial community. Iterative highest node will be assigned to the local community similarity in sequence to the community. By border measuer function Q determins algorithm termination condition. Automatically end the cycle. With comparing experiments by the algorithm to this article, GN algorithm and CNM algorithm, we believe The proposed algorithm has a better classification accuracy in conclusion.