Community Detection And Semantic Analysis In Attribute Networks

It is of great significance to identify the community structure in the network by using the complex network analysis method to analyze the relationship structure and attributes in the complex system.The research group has built an intelligence analysis framework IAF based on complex network algorithm for the perception,understanding and prediction of people,objects,organizations and their interrelationships.The detection of communities from large-scale social association networks(complex networks)based on human electronic footprints in social/physical/cyberspace is a key function of community detection in the understanding module of intelligent intelligence analysis framework.This paper proposes community detection methods combining network structure and node attributes,and makes an empirical study.Firstly,the community detection method(PLAC)combined with scale-free property of attribute network is proposed.Based on the joint generation model,this method preserves the heterogeneity of node degree in the network by encoding the degree distribution of changes in all nodes.The model combines the structure and attribute information of the network simultaneously for community detection and semantic identification.Secondly,the generalized community detection and semantic recognition method(GSC)in attribute networks is proposed.This method is based on the joint Bayesian generation model,the structure and properties of the network are modeled simultaneously,and the effective Gibbs sampling is used to solve the problem.The Bayesian mixture model is used to detect the generalized communities in the network,and a probabilistic transition matrix is used to reveal the potential relationship between the structured communities and attribute topics.The semantics of communities are interpreted by analyzing the content information of the nodes in the communities.Finally,this paper uses the paper data set for empirical analysis.According to the citation relationship of biological papers,an attribute network is constructed and the nodes are divided into communities.Based on the intelligence task of analyzing the status quo of field research,this paper also constructs a cooperative network of authors in different fields of physics to do empirical research.This paper explains the semantics of communities,analyzes the research contents in different fields,and provides guidance for future research activities in various fields.Based on the function of community detection in the understanding module of intelligence analysis framework IAF,this paper proposes two community detection generation models in the attribute network,and verifies it on the real data set.By using the content information of nodes,it analyzes the relationship between the community and the attribute topic,and describes the semantics of the community.At the end of the paper,an empirical analysis is made on the Aminer data set,which can discover specific communities and reveal their characteristics.It is of great value to analyze the information about the research hotspots in various fields.