Analysis Of The Network Structure In Social Network

Social network is an information exchange network of individuals and organizations, as well as their connections. In recent years, as the rapid development of information technology, Social network has gradually become the hot spot of the new interdisciplinary research.The network structure is an important feature of the social network, network structure not only reflects the regional characteristics of the individual node behavior in the network, but also reflects the important relationship between groups. According to the analysis of these basic elements which got by the analysis of the intrinsic relation between these data to mine the network structure, and then we get the hierarchical division of the network and division of community based on the network structure. The network structure can be used to analysis more profoundly and comprehensively the network robustness, survivability and stability.In this paper, the research of the characteristic of the social network structure is from two aspects which are the hierarchical structure and community structure. This paper also proposes structure discovery algorithms respectively in view of a highly organized structure of social network and the general structure of the social network. The first algorithm named organizational structure discovery based on the cut set and hierarchical method proposes a new measure of node importance which is suitable for use in highly organized network, and it combines the global information and the local information of nodes. It uses cut set algorithm to obtain the global information with the network node’s level, and the local information directly is obtained by the node’s output/input degree. Furthermore by using BP neural network method to optimize the integrated measurement parameters weights so that the accuracy of the measurement will be better, and finally it digs out the network hierarchy. The second algorithm named community detection based on fixed length random walks is a suitable method to find the general structure of the social network. The method based on the random walk model determine the steps which the number of iterations that transition probability matrix by using an error function, so that the algorithm does not need to consider the convergence problem of walking. The algorithm uses node similarity function and community evaluation function for clustering the network community. Comprehensive experiments conducted on real data prove the validity of the algorithm.