Cliques in complex networks reveal link formation and community evolution

Abstract: Missing link prediction in indirected and un-weighted network is an open and challenge problem which has been studied intensively in recent years. In this paper, we studied the relationships between community structure and link formation and proposed a Fast Block probabilistic Model(FBM). In accordance with the experiments on four real world networks, we have yielded very good accuracy of missing link prediction and huge improvement in computing efficiency compared to conventional methods. By analyzing the mechanism of link formation, we also discovered that clique structure plays a significant role to help us understand how links grow in communities. Therefore, we summarized three principles which are proved to be able to well explain the mechanism of link formation and network evolution from the theory of graph topology.