Research And Implementation Of Key Network Community Discovery Algorithm For Large-scale Academic Papers

In recent years,complex networks have become more and more active in scientific research.With the in-depth study of the nature of complex networks,people have gradually discovered some common properties in complex networks,such as scale-free characteristics,community structure and so on.The community structure in a complex network can be seen as a mapping of relationships in the real world.For example,in a network of academic cooperation,people who study common topics or compare similar topics will also form a group.In the academic paper keyword network,a relatively close technology or method will form a technical community.At this time,how to quickly and effectively find a better community structure in a large complex network has become a concern of researchers.Community discovery research is produced in this context.This topic is devoted to extracting keywords from a large number of scientific research articles,filtering the proposed keywords,constructing a large-scale keyword complex network,and then using the technology relevance analysis method to hierarchically divide the technology into a hierarchical structure.The technical community system explores the relevant subdivisions of the technology,including keyword information such as technology points,products,and applications.Based on the above,the main work of this topic is as follows:1.In response to the problem that the keyword set proposed by TextRank algorithm contains impurity words,this paper proposes KGCNN algorithm to extract and filter out some impurity words.2.In view of the fact that the traditional label propagation algorithm(LPA)algorithm does not consider the weight and the semantic relevance of nodes in the label propagation process,we propose an ELPA algorithm that considers the weight and node semantic correlation.3.Complete the design and development of the central enterprise technology innovation resource management service platform module.The system mainly selects and filters the acquired data(platform existing or custom crawl)based on the research field keywords provided by the user to construct a complex network.Finally,A series of algorithm analysis is carried out on the constructed complex network,and the collected information is processed in a deeper level such as roughing and refining,and then the user is screened for reliable and practical information.In summary,this thesis studies and proposes that KGCNN algorithm is used to extract and filter keywords,and based on LPA,an ELPA algorithm considering weights and node semantics is proposed to discover communities in complex networks,and the above algorithm is proposed.Applied to the system,achieved good results.