Local Community Discovery Based On Core Node And Relevant Study

The research of complex network was got attention by every field researchers gradually; network life makes the study become a hot spot, including the identification of core nodes in complex networks, local community detection and global network partitioning are becoming the core of the study of complex networks. Considering the shortcomings of the current community discovery algorithm, this paper will focus on algorithm suited for regular network, scale and large-scale network, including the global community discovery and local community found that only use local information to get the local community and the global community. In this paper, research work mainly includes:(1) This paper proposes a core node identification method based on node relevance. The evolution of community is based on one or more nodes spread in the network society, which reflects the core role of the community. Use cohesion of node that relevancy to identify the core node of the node. Correlate with the difference of node clustering coefficient and outer clustering coefficient to describe relevance of nodes. Experiment with the core node based on close degree method comparing with the results in a variety of classic in the network, which have verified the effectiveness of the proposed method.(2) This paper proposes a model based on Bayesian posterior local community discovery method. In view of issues of community detection with Newman modularity, a Bayesian posterior model based modularity (BS modularity) is inferred which combines the Newman modularity and nodes’recommending probabilities. A corresponding local community detection algorithm using the framework of adjacency merges is proposed. The algorithm can search the local community efficiently without the defects of the Newman modularity, which has problems on discrimination in sparse network and resolution on community structure. Comparing experiments on benchmark data validate the BS modularity.(3) This paper builds a global community discovery method based on the core node group. In order to adapt to large-scale network association found that found in this paper, considering the local community has the characteristics of lower space and time consumption, the framework of this idea is introduced into global network partition, a new algorithm is composed to detect global network society. This algorithm only use local information to find out the global network society, for big data network reduces the club found time effectively. The algorithm starts at founding the core node group, then coarse community classification and detailed partition to achieve the purpose of looking for community. The algorithm is compared with classical algorithm of fast Newman, after classic network and found that the algorithm is suitable for large-scale network.