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

The Community Structure of network were presented out as an essential feature of complex networks with the deepening study. Overlapping and Hierarchical Structure of network intertwined together very closely. It is not appropriate if we just study Hierarchical Structure ignoring the Overlapping, or study the Overlapping ignoring Hierarchical Structure.Community detection algorithm has been widespread concern in academic circles, but the overlapping community discovery algorithm, especially overlapping community detection algorithm research using link-community is relatively new research field, which has a very broad prospect of research and important practical significance.Spectral analysis method is applied in link-community discovery algorithm for the first time the method, to achieve overlapping community discovery. The organic combination of both the above, one can improve the spectral analysis method for overlapping community discovery; the other spectral analysis can provide a new method for the link similarity quantitation in the network.Based on line-graph standard Laplacian matrix and line-graph normalized Laplacian matrix spectral analysis, taking into account the community hierarchical structure and overlapping, detect complex network overlapping community. This paper chooses to test on Zachary karate network and Dolphin social network, compares with and analyses Ahn’s algorithm which presented in 2010. Experiment results show that this algorithm can realize the complex network division of overlapping communities, and with better performance compared with Ahn’s algorithm.