The Weibo User Community Detection Based On The Combination Of Link And Content

Many systems in the real world can be abstracted as networks,such as social networks,essay citation networks,cooperation networks of scientists,and weibo user relations networks.These networks have the same characteristics: the complex internal structure,so called complex network.Some researches have shown that these networks contain some potential community structures,which are characterized by densely linked nodes within the community and sparse links among the communities.In general,nodes in a community have similar characteristics and play a similar role in the network.Identifying the community structure in the network through community division helps people to understand the nature of the network in more depth and to understand the relationship between the network structure and its functions.However,traditional complex network community partitioning algorithms generally lack comprehensive consideration of link structure and node content.Most of the methods of community division that combine the existing link structure and node content are based on probabilistic model.Such methods have the advantages of beautiful mathematical forms and strong interpretability,but they also have the disadvantages of high time complexity,difficult to understand and difficult to implement.Therefore,the purpose of this paper is to find a way to considering node content and network structure synthetically so as to obtain similar content and compact community.Based on the research on the method of obtaining the data of Weibo website,this paper first analyzes the user behavior characteristics in the social network and builds the user influence evaluation model to obtain the "core users" in the network.Based on the classification of initial influence structure,clustering work based on the similarity of user topics is carried out,and the potential semantic information behind the texts is excavated based on co-occurrence and implicit semantic analysis,and the dimensionality reduction through feature space is more accurate Weibo community division.It combines several excellent initial nodeselection methods and effectively implements the community division of content networks containing attribute information.Subsequently,the results of the initial division of community mergers,reducing the number of small communities,community access to more value of the community structure,the rationality of the algorithm and the complexity of the analysis.Finally,on the modeling of high-link community partitioning with content,it is ensured that the content of users in the community is of similar interest and structurally close.The method proposed in this paper is applied to a real dataset.Experiments show that the proposed method not only can identify potential communities,but also can learn community topics and solve the problem of the lack of semantic interpretation of the traditional community discovery methods based on the link structure.In addition,the research results are summarized,and the next research work is expected.