Researching On Communities Detection Algorithm Based On Edges Density Of Complex Networks

Recently, many scientists and scholars pay more attentions on the researching complex networks along with the rapidly developing of information technology. And many structural characteristics of complex networks put forward successively. However, community structures are a common structural characteristics of complex networks. In addition, overlapping communities are very common in real-world complex networks which have many vertices with different types. So accurately detecting overlapping communities is a basic and important task to understand the structures in real-world complex networks. Consequently, detecting community structures in complex network based on the number of edges between vertices, the main work and results include the following aspects.(1) An algorithm is proposed for detecting overlapping communities, which is based on the clustering threshold. Firstly, the characteristics of maximal complete sub-graphs are introduced. At the same time, the clustering coefficient between sub-graphs are proposed. This algorithm is that the maximal complete sub-graphs are extracted from complex network firstly. Then some maximal complete sub-graphs are gradually merged according to the clustering coefficient between the maximal complete sub-graphs. Lastly, different overlapping community structures are detected by setting different threshold value of the clustering coefficient. The results of experiment express that this algorithm based on the clustering threshold can accurately detect overlapping community structures with low time complexity.(2) An algorithm is proposed for detecting communities, which is based on the maximal complete sub-graphs and the intimate degree. So, connecting vertices, isolated vertices and intimate degree are introduced. Firstly, the maximal complete sub-graphs and isolated vertices are extracts from complex networks in this algorithm. Then the intimate degrees are calculated, which are between the vertices and the maximal complete sub-graphs or between the maximal complete sub-graphs. Some maximal complete sub-graphs are gradually merged by some rules given. The results of experiment express that this algorithm can detect overlapping community structures and some vertices with special properties with low time complexity.(3) An algorithm is proposed for detecting overlapping bi-communities from the bipartite networks, which is based on the initial key bipartite sub-graphs and the polymerization degree of bipartite vertices. The results of experiment express that the algorithm can accurately explore the overlapping bi-community structures.(4) In order to analyze the relationship of the same type nodes in the bipartite network, the original binary network is projected into two weighted one-mode network through the binary clustering triples. One-mode community structures are detected from the two weighted one-mode network by weighted clustering threshold. The results of experiment express that the algorithm can explore the overlapping one-mode community structures.