Research Of Semantic Community Detection Based On Quantizing Progress

With the development of network social communication technology,the online social networks,such as Weixin,Weibo and Facebook,have become indispensable social platforms in daily life.In these platforms,people generate community structure according to similar interests,hobbies or common topics,which satisfies the close links within the community and the sparse links between communities.Accurate identification of detected community structure in social networks is a complex task.On the one hand,it is easy to fall into local optimum solution to solve the topological correlation between nodes from a global perspective.On the other hand,nodes may have complex semantic information,such as posts in micro-blog,circles of friends in WeChat.These semantic information bring great challenges to community detection.This paper mainly solves the task of semantic community detection.Specifically,to solve the problem of quantification of semantic information and overlapping of community structure,it is necessary to establish a quantitative mapping process from node semantic information to semantic space with a large number of nodes and complex semantic information.Taking the semantic similarity as a parameter,then a new strategy semantic approach is proposed.Finally,a new modularity model of measurable semantic social network community is proposed.This paper mainly completes the following aspects:1.Construct the probability distribution of text information.LDA(Latent Dirichlet Allocation)model is used to plan and analyze documents,topics and keywords in text information,and the required information is extracted as the semantic information model of nodes.2.Solve the implicit parameters in LDA model.Compared with EM method and expectation push method,Gibbs sampling method is used to estimate and analyze the uncertainty accurately,thus completing the quantitative mappingfrom semantic information to semantic space.3.Constructing particle swarm strategy and modularity measuring function of semantic community.The network node is the basic particles in particle swarm optimization,and the connection probability between nodes is analogous to the speed of particle walking.A PSO(Particle Swarm Optimization)strategy with semantic relationship is proposed to solve the problem of community detection.According to the projection angle of nodes in two-dimensional space,a modular measure function(Sim Q)based on semantic community is constructed to evaluate community quality.4.Experimental results and analysis.Through experiments,the performance differences between PSO-LDA and various community detection algorithms in synthetic and real networks are verified by the number of topics in semantic communities,intelligent optimization strategies and semantic modularity.The experimental results verify the effectiveness and feasibility of the algorithm.