Learning community structures

摘要: Uncovering community structures is a fundamental and important problem for analyzing complex networks. The topology information, as the direct representation of networks, widely used detection. But in fact, there are other two types information related with network topology: global which captures importance nodes whole network, local describes similarities between nodes. It great value to consider individual them detection methods simultaneously, largely ignored by previous methods. In this work, we integrate uniformly novel nonnegative matrix factorization (NMF) based model. Specifically, aspect, employ PageRank derive nodes, so that more node is, influence network. utilize nearness obtain larger will have similar memberships. Thereafter, multiplicative updating rule learn model parameter. Numerous experiments demonstrate our approach has gained performance improvements up almost 5% comparison state-of-the-art