Community detection has attracted a great deal of attention in recent years. A parsimony criterion for detecting this structure means that as minimal as possible number of inserted and deleted edges is needed when we ...Community detection has attracted a great deal of attention in recent years. A parsimony criterion for detecting this structure means that as minimal as possible number of inserted and deleted edges is needed when we make the network considered become a disjoint union of cliques. However, many small groups of nodes are obtained by directly using this criterion to some networks especially for sparse ones. In this paper we propose a weighted parsimony model in which a weight coefficient is introduced to balance the inserted and deleted edges to ensure the obtained subgraphs to be reasonable communities. Some benchmark testing examples are used to validate the effectiveness of the proposed method. It is interesting that the weight here can be determined only by the topological features of the network. Meanwhile we make some comparison of our model with maximizing modularity Q and modularity density D on some of the benchmark networks, although sometimes too many or a little less numbers of communities are obtained with Q or D, a proper number of communities are detected with the weighted model. All the computational results confirm its capability for community detection for the small or middle size networks.展开更多
基金This research is partially supported by the National Natural Science Foundation of China under Grant No. 60873205, Innovation Project of Chinese Academy of Sciences, kjcsyw-sT.
文摘Community detection has attracted a great deal of attention in recent years. A parsimony criterion for detecting this structure means that as minimal as possible number of inserted and deleted edges is needed when we make the network considered become a disjoint union of cliques. However, many small groups of nodes are obtained by directly using this criterion to some networks especially for sparse ones. In this paper we propose a weighted parsimony model in which a weight coefficient is introduced to balance the inserted and deleted edges to ensure the obtained subgraphs to be reasonable communities. Some benchmark testing examples are used to validate the effectiveness of the proposed method. It is interesting that the weight here can be determined only by the topological features of the network. Meanwhile we make some comparison of our model with maximizing modularity Q and modularity density D on some of the benchmark networks, although sometimes too many or a little less numbers of communities are obtained with Q or D, a proper number of communities are detected with the weighted model. All the computational results confirm its capability for community detection for the small or middle size networks.
