Abstract: Fast Unfolding is a hierarchical community detection algorithm based on modularity. It runs very fast, but the accuracy needs to be improved. Because the algorithm adopts traditional modularity to merger communities, it only considers node link information and ignores the neighbor nodes. Therefore, two nodes that have common neighbors and weak link information may not be merged, thus affecting the accuracy. In view of the shortcomings, a hierarchical agglomerative community detection algorithm based on similarity modularity was proposed through introducing optimized similarity to improve the modularity. It adopts NMI as the accuracy measurement. Experiments on the real network and LFR synthetic network show that the accuracy of detecting community is obviously improved.

Key words: fast unfolding algorithm, modularity, similarity, community detection