Research On Algorithms Of Overlapping Community Detection In Complex Networks Based On Line-graph

In recent years, overlapping community detection algorithm based on line graph becomes a rising new field and has broad prospect. Line graph is a method that treats the community as a partition of the links rather than the set of nodes, whose advantage is to use non-overlapping community detection algorithm as an overlapping community detection algorithm. In this paper, based on the line graph, we propose an overlapping community detection algorithm.In the real world, the number of communities of complex networks is unknown, which makes some algorithms that depend on the priori knowledge of the number of community cannot be used. In this paper, fisrly, we applied the method based on Jordan Matrix of Laplacian Matrix to line graph to obtain the number of community of line graph. Secondly, we applied the spectral clustering algorithm to line graph, and map the links to Euclidean space by eigenvectors. The component elements of eigenvectors correspond to nodes of line graph. Then select two eigenvectors clustering sample and calculate their similarity. Lastly, with the prior knowledge of the communities, use the classical clustering algorithm K-means to detect overlapping communities, which take advantage of K-means algorithm. On the other hand, apply hierarchical clustering algorithm to line graph and use the priori knowledge obtained to cut hierarchical clustering tree, which reveal the overlap and hierarchy of complex network simultaneously. Experimental results show that the method in this paper can detect overlapping communities and has better performance than related algorithms.