The Research On Community Structure Analysis Methods Of Complex Networks Based On Relationship Strength

The unprecedented development of information technology keeps promoting the analysis of complex networks. The reveals of many important characteristics of complex networks, such as "scale-free" and "small-world", are of great significance to the understanding of networks’ structure and evolution. Community detection has attracted much more attention because of many difficulties such as the complexity, the overlap of the community, and the large scale of the networks.This paper starts the research on the related issues of community detection, and the details are as follows:1) The metric of relationship strength is proposed. This metric defines the closeness of arbitrary two nodes, which is similar to the similarity. And all the proposed algorithms are based on the metric.2) A community detection algorithm based on core-clique is proposed. The concept of core-clique is proposed according to the characteristics of complex networks, based on which a non-overlapping community detection algorithm is proposed. The main procedure of the algorithm is adjusting the distribution of the core-clique by a pre-allocated method. And then, an overlapping condition is proposed based on the extended modularity. An overlapping community detection algorithm is proposed based on the non-overlapping community detection. The experiments show that both algorithms could mine the community structure accurately.3) A local community detection algorithm is proposed. A new fitness function is proposed based on average precision and the relationship strength, based on which an expanding model is designed and a new local community detection algorithm is proposed. The experiments show that the algorithm is fast and efficient.4) A crawler for network of weibo is designed. A local weibo network is crawled by the designed crawler, on which the community structure is detected by the proposed local algorithm. And a verification model is designed for community detection. The results show that weibo network has a significant community structure.