Research On Community Detection Algorithms In Weighted Complex Networks

In recent years, with the development of Internet and computer technology rapidly, especially the increasing penetration of Web2.0 era, the scale of complex networks which can express the relationship between somewhat shows a rapid growth trend. Research on complex networks has become a hot issue of common concern of academia and industry. Complex networks not only has the characteristics of small world, no scaling, but also has the obvious characteristics of the community structure, thus community detection in complex networks can solve many specific problems of practical applications. However, in real life, many relationships in complex networks are not the simple existence of Boolean, that is to say, there are differences existing such as, distances between nodes, level of rights in complex networks namely the weighted complex networks. Therefore, research on community detection in weighted complex networks has more and more practical significance and research value.This paper is aimed at the study of complex networks in reality, especially in the weighted complex networks, using the method of community detection. First of all, according to the characteristics of weighted complex networks, combined with the label propagation principle, puts forward the weighted complex networks community detection based on label propagation. The algorithm by adjusting labels of nodes constantly, deletes the edge which has the maximum weight, and ultimately achieves convergence, to find communities in weighted complex networks rapidly.Secondly, according to the characteristics of weighted complex networks, the concept and application of edge-betweenness in Girvan-Newman algorithm, puts forward a new concept of Relative-betweenness. Deleting edges which have the minimum number of Relative-betweenness, get up connected sub-graph, and ultimately achieve the convergence condition. With community detection in weighted complex networks, the algorithm improves the effectiveness.At last, compared with GN algorithm and proposed algorithms in this paper using three kinds of weighted complex data, the performance of the community detection algorithms is satisfying.