A new community detection based on agglomeration mechanism

Abstract

Community detection is an important and hot research branch of complex network. The initial communities are essential for community detection, and the central nodes are the key points in the whole process. A few network measures are employed for node centrality, including betweenness and degrees centrality calculation. In our proposed algorithm both methods will be tested respectively for initial communities. Moreover, the agglomeration mechanism is employed for the proposed algorithm, and corresponding communities are achieved according to node membership function. Communities will be merged repeatedly based on the communities agglomeration rule until the defined number of communities is achieved. The proposed algorithm is tested on the three real network datasets, and it demonstrates the effectiveness and correctness of the algorithm.