Author(s)

Xiaoyuan Xu

Corresponding Author

Xiaoyuan Xu

Abstract

Community detection is the basis of studying complex network structure. Based on the analysis of existing overlapping community detection algorithms, an edge-based overlapping community detection algorithm SAEC is proposed. The algorithm regards the community as a set of edges. By defining the similarity of edges, the probability transition matrix is obtained. The number of communities is automatically determined by spectral clustering method. Finally, overlapping communities are divided by K-means algorithm. The validity of the algorithm is verified by the test of random generated network and real network.

Keywords

Network community; spectral clustering; edge; detection