关键词: 复杂网络, 社团检测, 极大团, 模块度矩阵

Abstract: Community structure detection is instrumental in revealing the structure-function properties of complex networks. The existing community detection algorithms suffer from resolution limitations, node uncertainty, and the need for prior parameters in their research process. A community detection algorithm based on the maximal clique of network nodes (BMC) is proposed to solve these problems. The BMC algorithm sets the maximal clique of nodes in the network as the initial node cluster, and merges the node clusters by hierarchical clustering based on the proposed local similarity of the maximal clique and the local clique relationship, so as to mine the community structure in the network. Aiming at tackling the issue of node uncertainty in the mining process of community structure, the module membership degree is proposed through the modularity matrix to optimize the single neighbor nodes and overlapping nodes in the network. In order to verify the accuracy of the BMC algorithm for network community structure mining, experiments are conducted on five real network datasets with five algorithms for comparison. The experimental results obtained by the three measures show that the BMC algorithm accurately detects the community structure in the network.

Key words: complex network, community detection, maximal clique, modularity matrix