Detecting Community Of Complex Network By Local Information

In recent years, the network community structure discovery algorithm research and application is becoming a hot spot of multidisciplinary joint research between. Although there has obtained certain achievements, but the time complexity and clustering quality and stability of the algorithm is still not satisfactory, and problems to be solved urgently are as follows: in a large amount of data and topological structure of complex networks improve the clustering precision of the algorithm of mining community structure; Nodes in the network overlap division effectively; On the premise of the lack of prior knowledge, the community structure of consistency of results.To solve above problems, on the basis of predecessors’ research, this paper put forward based on the greedy expansion of overlapping community partition algorithm; In view of the overlapping community structure network, based on very greedy expansion of community discovery algorithm was proposed. We adopt standard virtual cube respectively comparing experiments, the main content of the thesis are as follows.First, this paper outlined the complex network of related knowledge, and emphatically introduces the research status of community discovery algorithm study, including the present typical community discovery algorithm, and their respective characteristics and the deficiency existing in the algorithm.Second, we found in relatively mature non-overlapping network community study, because of the large amount of network data and complex community structure resulting in a decline in the clustering accuracy of the algorithm, the algorithm divided societies inconsistent results. In this paper, based on local information algorithm is proposed on the basis of the mass of the division of greedy expansion algorithm, the algorithm USES the regiment seeds do algorithm, effectively improve the clustering algorithm.Then, in view of the complex overlapping communities overlap nodes in the network is not easy to divide the problem, this article uses the edge as a measurement tool, the greedy expansion algorithm based on local information above research work, by sifting through the network to gather degree of high side edge, and as the seed of community discovery algorithm was proposed based on super seeds of greedy expansion algorithm, clustering precision of the algorithm.Finally, this paper respectively in a complex network of standard data sets on the proposed two algorithms to do the experiment, and comparing and several classical algorithms, and the algorithm of clustering accuracy and stability is analyzed.