Social Network Community Discovery Research For Weibo Users

The social network changed the way of communication which transform the new era of "online" and "palm" communication instead of the traditional offline way with the rapid development of computer technology.Now the huge amounts of data in social network are vital for social science research with the coming of the era of big data.The community structure of social networks is becoming a hot spot of scholars study field.The community detection technology has great help to the study of complex network topology,and also contains a lot of value for society.At present the community detection technology has made great progress in the complex networks field,but it is not very mature,because of the scale large and content complicated for social network.Most algorithms have some defects,such as the algorithm complexity is too high,the result is inaccurately or exists local optimum.In view of this,this paper researches on the microblogging network of the social network,starts from the microblog users and through the integration of the relationship between the user and the user content to find the potential user community,and it is verify the rationality of the results through the experiment.This paper mainly study from the following aspects:(1)According to characteristics of users with each other in the microblogging network structure,there are two classes of users' relations about two-way and one-way attentions,this paper proposes a method of users' relationship similarity.The method considers the influence of the two kinds of concerns on the nodes,and converts the network into a weighted undirected network to improve the efficiency of the operation.After that,this paper proposed an improved CNM community discovery algorithm based on the user relationship similarity.According to the ideas of the friends of friends are more likely to become the friend,it can be extended to friends of friends with their similar,so using alternative module node similarity degree of community merged,made user community discovery more reasonable.This is community discovery based on the characteristics of the relationship between the users in the network.(2)The user's contents can reflect their current interests in microblogging network,according to this idea,it proposes a community discovery algorithm fusing the relationship between the user's relationship and user's contents.It uses the topic model to find the user interests topic with user tags,and calculates users interest similarity by relative entropy,meanwhile adds the user similarity relations and adjusts the proportion of two types of similarity fusion through the experiment.It fully reflects the user's interest.(3)This paper proposes a community discovery algorithm named JSCNM algorithm based on the integration of similarity measure computation with the relationship between users' relationship and users' interests.It is carried out on the relationship between user and user interest merged degrees into the optimization function by improving the module degrees,after constantly search for the optimal partition to found the user community.The results prove that divide the community more reasonable by use microblogging network real data set through the experiment.