Research On Community Detection Based On User Influence In Microblogging

In recent years, the rapid development of microblogging has caused more and morepeople’s attention. Microblogging emerge from the traditional social network, and it has theindependent service platform which has gradually evolved into a new form of informationrelease. But in the rapid development it also faces some problems at the same time, such as fraudinformation, false information, rumors, slander or even reactionary information which can spreadquickly through the microblogging, so much as triggering malignant mass incidents. Supportedby the national863plan project,“Common Security and Control Framework in Tri-NetworkConvergence” project. Based on the complex network research, this paper takes a research oncommunity detection in microblogging. We propose a new user influence evaluation model. Andcombined the model with community detection algorithm, it realizes the division of themicroblogging community. Through the study of this paper, it is particularly necessary tocommunicate effectively and efficiently, and it has great significance on marketing management,supervision and guidance of public sentiment in microblogging. Its main work and contributionsare outlined as follows:1) By using Sina Microblogging site to get real-time data, we make a statistical analysis ofuser’s behavior characteristics. Based on this, it advances a kind of user influence evaluationmodel. The model realizes the importance of microblogging users in the network assessment.Experimental results show that the model is a more accurate evaluation of the influence ofmicroblogging users.2) Studies have found that there are a large number of stellate sub-graphs in microblogging.Combined with the results of user’s influence, we propose a community detection algorithmbased on stellate sub-graphs. This paper theoretically analyzes the rationality of the algorithmand the algorithm complexity. And simulation results show that by using this algorithm canobtain better community detection effect. At the same time the algorithm is improved in thealgorithm’s execution efficiency.3) By using the ideas of Newman’s proposed module functions, this paper presents acommunity detection algorithm based on seeds expansion. Algorithm fully considered thecharacteristics of the node attributes. This algorithm makes up for the existing algorithms onlyby using the connection between nodes to divide community respectively. Experimental resultsshow that the algorithm is more close to the actual situation in the community division results.But because of the added node properties, it increases the algorithm complexity. Comprehensiveconsidering various factors, in order to meet the requirement of real-time and effectiveness whenanalyze the microblogging network, this algorithm can be combined with community detection algorithm based on stellate sub-graphs to divide community. On the basis of the latter quicklyidentify achieve more accurate classification of the entire network.