Research On The Construction And Clustering Of Bibliographic Semantic Coupling Network

With the rapid development of network information technology,the field of text mining has been paid more and more attention,and has become one of the hotspots of current research.Text mining refers to the acquisition of valuable information and knowledge from text data,one of the most commonly used methods is to use bibliographic coupling to carry out retrospective analysis of the existing literature.The traditional bibliographic coupling network is constructed by the coupling intensity between literatures,which only considers the coupling relation,and does not consider the similarity between the contents of the literature.In order to depict the similarity relationship between literatures more accurately,this paper adds semantic information on the basis of bibliographic coupling network,constructs the bibliographic semantic coupling network,and carries on the clustering analysis to the constructed network.The research work of this paper is mainly carried out from the following three aspects:Firstly,constructing the bibliographic semantic coupling network.Through the LDA modeling method,the semantic information of nodes in the network is quantified,the semantic information of nodes is combined with the coupling relationship between nodes,the semantic characteristics of the network are considered on the basis of network topology features,and the semantic field model of semantic network is constructed by the idea of social network topology potential,and then the semantic coupling network of literature is made.Secondly,the evaluation index of modularity is optimized.Because the bibliographic semantic coupling network constructed in this paper not only considers the coupling relationship between nodes,but also considers the semantic information of nodes.Therefore,the corresponding community detection evaluation index should consider not only the rationality of the relationship within the community,but also the similarity of semantic information between nodes.Based on the above considerations,a semanticmodularity evaluation index is defined.Through comparative analysis,the superiority of semantic modularity evaluation index is verified.Finally,cluster analysis and community detection are carried out on the network model.Using GN,LPA and Louvain algorithm,this paper makes community detection to the bibliographic semantic coupling network,and analyzes the research topics of different communities according to the result of the division of the network and the keyword information contained in each community.At the same time,the Visual feature analysis of bibliographic semantic coupling network is carried out,and the internal law,research hotspot and discipline structure of the network are excavated.The experimental results show that it is found that after considering the literature content on the basis of literature coupling,the connection between nodes is closer,and the development dynamics and research trend of the subject can be mastered accurately.In this paper,the improved bibliographic semantic coupling network and Semantic modularity index provide a new research idea for the research literature citation structure and law,subject similarity and discipline structure,which has certain reference significance for literature clustering and information retrieval.