Research On Key Structure Of Large Scale Socialnetworks Based On Stochastic Block Model

With the rapid development of the Internet, widespread use of smart phone,pad and other smart terminals, and the improvement of mobile networkbandwidth, social media such as wechat, microblog and etc. has been becominga significant part of human life. As the people’s interactions and relationsextension to the internet, our society has been networked either. In this situation,as a cross hot of sociological theory, information processing technology andelectronic commerce, large scale social network has come into being. Mass datawhich contains a lot of useful information has been generated in large scalesocial networks every day. Detect and analysis these data has already been anessential tool to study the structure of large scale social networks. We canunderstand the large scale social networks better and provide a theoreticalfoundation to practical application such as public opinion monitoring,personalized recommendation in e-commerce and etc. by analysis and detectuseful information of large scale social networks data.Currently, the researches of social networks are no longer concerned aboutsmall and medium-size social networks. Instead, research increasinglyconcerned about the large scale social networks. Meanwhile, the target of research has transformed into multi structures from single structure. Theresearch on large scale network multi structures has been a popular study pointof social networks structure. However, the existing methods only concernedabout the determinate structure instead of the combination of multi structurewhich can express the structure of large scale social networks better. Based onthese considerations above, this paper proposed a fresh notion, the definition andthe practical algorithm after study the problem of large scale social networksstructure. The main contributions of this essay include the following aspects:First of all, based on the traditional definition of large scale social networks,after take full account of the consequence of multi structure in structure analysisof large scale social networks, we proposed a notion named key structure. Baseon the definition, we illustrate the detection process and provide a validevaluation criterion. Secondly, we proposed a novel definition of structural holespanners and introduced the stochastic block models SBM and the method ofstructure hole spanners detection SHSD respectively which will be used in keystructure detection. Finally, we did a series of experiments by using a blogdataset named BC dataset and a microblog dataset which we named MB datasetas the experimental datasets. The simulation results show that the stochasticblock model can detect the community structure of the social networks and theclassification results can be used in detect the structural hole spanners in thenetwork by SHSD. Comparing with the original network data graph, the detected key structure can full describe the structure regularity of socialnetworks.