Research Of Community Detection Based On Complex Networks

With rapid development of modern science and technology, the amount of information is growing exponentially. However, processing mass data are becoming the bottleneck for people to get valid information. With the improving circle of friends, detecting the community structure in complex networks has important theoretical significance for analyzing the property, function, regular pattern of complex networks and forecasting the behavior of complex networks.There are a large number of topological features which is invisible in simple complex networks. The research history of complex network is not long but is a top field in science research. In most cases, the features of complex networks are studied by researchers through observing the phenomenon of real world or computer network such as www networks and social networks. The main tasks in this paper can be listed as followings.1. An algorithm of mining community by using local information based on CNM algorithm is proposed. Firstly, the vertex's intensity and contribution for community is defined as the measure of calculating the similarity of data of same class in clustering. Some modification is made to the definition of weighted modularity so as to obtain the adaptability for both weighted networks. Several small communities are obtained utilizing local information of communities. Secondly, the set of small communities is used for the input of CNM algorithm. The final result is obtained through computing the increment of the modularity and clustering the small communities with little increment. The experiments show the slight improvement on both the modularity and executions time of the proposed algorithm.2. A community detection algorithm based on Jaccard similarity is presented. Firstly, the similarity of the vertices is used for finding the near neighbors. Secondly, the best neighbor is selected from the neighbors. Finally, the small communities consisting of several vertices are formed by merging two vertices who are friends. Furthermore, the CNM algorithm is used for clustering small communities to complete the community detection cluster. The increase of modularity for the detected communities and decrease of the execution time shown by the experiments prove the effectiveness of the proposed community detection algorithm.In recent years, community detection has been a hot research field. People get lots of benefits by community detection in livelihood. In future, the focus of research topic should be shifted to use the sophisticated algorithms for serve for our production and life. The main method will be find suitable modularity function to dividing communities.