Detecting an Overlapping Community Structure by Using Clique-to-Clique Similarity based Label Propagation

摘要

Many researchers have proven that complex networks have community structures and that most network communities are overlapping. Numerous algorithms have been proposed and used to detect non-overlapping or overlapping communities in networks. Many community-detecting algorithms are based on a clique. A clique is a subset of the nodes in the network in which every pair of nodes has an edge between them. In this paper, we propose a new algorithm that is based on a clique-to-clique similarity measure, and the label propagation to detect overlapping communities. The algorithm first finds all cliques of the network; then, it builds a new network according to a specific strategy, that specifies that in the new network, a node represents a clique found in the last step, and an edge is the link relation generated according to the strategy. The experimental results for both synthetic networks and real-world networks show that the proposed algorithm is not only effective, but also better than other algorithms in forms of the quality of results on the time efficiency.