Research On The Local Community Detection Algorithm Of Social Network

Social network refers to the permanent, stable relational schema of society between people and people, collective and collective or individual and collective. Research shows that social network has a strong community features, the research of social network community structure is conducive to a better understanding of the social structure. In the current, social network under the condition of increasing in scale and frequently updated, global community detection become more and more difficult, the method of local community detection is becoming more and more important and has extensive application value. The main problem local community detection solve is that: a local community, a node or a few nodes located, will be quickly found within the local scope, without the high cost of time and space in global calculation. This article mainly aims at the current problems existing in the local community detection algorithm, some improvement research is conducted. The concrete content is as follows:Firstly, Due to the choice of the initial node, the calculation result of some existing algorithms is not robust, reasonable node search strategy can be very good to improve the result of local community detection algorithms. In order to solve this problem, this paper proposes a local community detection algorithm based on selection probability. The algorithm only gathers one node every time, according to the incremental value of the module, it calculates the selection probability for every candidate node, to make the nodes with high module incremental have high selection probability, avoiding local optimum because of the choice of a node with maximum module value every time. Experiments show that the algorithm can effectively find local community, and it is superior to the other algorithms.Secondly, Most algorithms begins with a node, the gather ability of a single node must be less than the multiple nodes, in order to solve this problem, this paper proposes a local community detection algorithm based on least cluster. According to the initial node the algorithm will firstly find the smallest clusters of closest connection containing the initial node, they must be in a community. Then it will detect the local community on the basis of the least clusters. Experiments are carried out on simulation networks and real networks, comparing the results of several algorithms, it show that this algorithm significantly improve the quality of the local community.