RETRACTED ARTICLE: A weak clique based multi objective genetic algorithm for overlapping community detection in complex networks

Abstract

Community detection is an important technique to find hidden information, pattern, and relation in a complex network. All the traditional community detection algorithms focus on the separated community but in real-world networks, most of the community is overlapping. Detecting overlapping community thus becomes essential. Although overlapping community detection techniques based on clique, has exposed hopeful performance but suffers from the serious curse of dimensionality due to its high computational complexity, when comes to large-scale networks. To deal with this drawback this paper proposes a weak clique based multi objective evolutionary algorithm for overlapping community detection. In this algorithm, a novel gene pattern is designed using a weak_clique graph. The weak cliques in the original graph are the nodes in the weak clique graph. The weak clique is formulated by selecting two neighbours which are very similar. Any two nodes in the weak clique are differed at most two node distances. The weak cliques comprise of more than one actual node thus weak cliques share actual nodes. This property of weak clique graph allows finding overlapping communities with high accuracy and low computational cost. Experiments on a variety of real-world and artificial networks validate the performance of the proposed algorithm.