Research On Community Detection Problems In Social Networks

With the development of social and technology, the communication of people also gradually frequently,especially some social networking sites and the rise of social software, make people more frequent close contact, then formed a large social network.Community structure is regarded as one of the most important properties of a social network.Communitydetection in the network community structure has important significancefor us to understand the structure and function of the social network. This paper mainly study community detection problem in the social network, the main contents include the following several aspects:Firstly, we research the traditional label propagation algorithm which regard the importance of all the nodes in the label propagation algorithm as the same, and has a lot of randomness which results the detected community in a low quality. This paper proposes a label propagation community detection algorithm based on intimacy. According to the importance of nodes, this paper gives the concept of local center node and assigns a unique label of these nodes. The algorithm can reduce a lot of unnecessary judgment during the process of transmission cost and improve efficiency. At the same time in the process of label update consider the closeness of nodeswhich can avoid the randomness in the process of nodes in the update label, so as to improve the stability and the quality of the algorithm.Secondly, at present the community detection method based on similarty only considering from the similarity between nodes.This paper puts forward the similarity of community detection algorithm based on improved the node similarity.The main idea is puts the edgeimportance between nodesinto the Jaccard similarity of nodes, and then presents the definition of the integrated similarity. Calculatethe community similarity among the nodes and put the nodes into community which have the larger community similarity.Experiments in real network and the artificial network show that our approcahhas a good community detection result.Finally, through the observation of the real network, people found that the importance of each node in the social networks are not the same, usually the greater influence of points in the community have more important centricity. This paper puts forward a kind of community detection algorithm based on the core node, first through computing the local influence of nodes to find the core nodes, and the second is detected the local core communityby the core nodes, last is to detect all the community structure of the network.This algorithm based on local method need't to know the number of communites before detection but has low time cost and works well.