Research On Hierarchical Community Structure Model Based On Fusion Similarity In Social Network

With the rapid development of information technology,social networks are becoming more and more important in people's life.In the real world,many systems can be expressed in the form of social networks,such as protein interaction networks,social networks,sensor networks,scientists collaboration networks and so on.It is hoped that through the analysis of the social networks,we dig out the hidden information,rules and knowledge.Among them,the problem of community detection is an important problem in the research of social networks,through the study of community detection,we can understand the structure and function of the social networks,find hidden patterns in the network and predict its behavior,it has very important significance to effectively understand social network.Therefore,there have been many research results in the past decade.But the current research work on the community detection mostly focused on the community division in the view of a single granularity,without considering the multi granularity and multi hierarchical community division.The existing work on the multi hierarchical community division have only take into account the structural relationship of the network,and ignoring the similarity relation of the vertex attributes and this problem is also found in the overlapping community discovery algorithms.For the community classification results have found,very few people to improve the quality and accuracy of the community division through further outlier detection and correction.Therefore,we present a hierarchical community division algorithm based on fusion similarity and the outlier detection and correction algorithm.The main contributions and work are as follows:(1)For the defect of hierarchical community division,which is rarely integrated structure and attribute information of the network,we make a detailed analysis and present a hierarchical community division algorithm based on fusion similarity,this algorithm considers both the similarity of structural between the communities and the attribute similarity between the vertices,and constructs the hierarchical structure of community division from bottom to top,so that the community division of the entire network is more scientific and accurate.(2)In order to find the overlapping vertices and improve the quality and accuracy of the community division,we present an outlier detection and correction algorithm,the algorithm is constructed in the community tree,and then we select a community division to dig out the outliers and overlapping vertices in the communities,then make them corrections respectively.