Research On The Evaluation Of Scholars’ Influence From The Perspective Of Social Networks

In recent years,with the continuous increase of scholars and scientific research output,academic evaluation has attracted more and more attention in various research fields.As an important part of academic evaluation,evaluating the influence of scholars is not only of great significance to guide the talent introduction,performance evaluation and scientific research fund allocation of colleges and institutions,it can also provide reference value for scholars to improve their personal academic influence,carry out scientific research in a more comprehensive way and publish high-quality papers.Existing studies have been devoted to exploring the factors related to the influence of scholars from different perspectivess,mainly based on the evaluation of a single factor.The method of integrating multiple factors for analysis from the perspective of network structure is still lacking.From the perspective of two-layer network structure,combined with various factors related to scholars’cooperation and paper citation,this work proposes a comprehensive evaluation method of scholars’influence.Firstly,based on the cooperation relationship between scholars and the citation relationship between papers,this work constructs the scholar cooperation network and paper citation network respectively.Using the affiliation relationship between scholars and papers,a two-layer coupling network model of scholar cooperation paper citation is constructed.Through the modeling method of presenting scholar cooperation and paper citation in a network,the joint effect of the two on the influence of scholars is comprehensively considered.Secondly,combined with the structural characteristics of the two-layer network,the indicators related to the influence of scholars from three aspects are extracted,the status of scholars in the network,the reputation of scholars and the social relations of scholars:the centrality of scholars in the cooperative network,the reputation indicator and the team size of scholars.Among them,scholars’reputation indicator realizes the coupling between the two-layer networks,reflecting the effect of paper citation on scholars’influence.Finally,using indicators extracted from three different angles,a scholar influence representation model(SIRM)is constructed,and the ability of the model to identify and predict high influence scholars is further tested.Through the analysis of Aminer data set and the comparison with other prediction models or ranking methods,the effectiveness of SIRM model in identifying and predicting high impact scholars has been proved.Three main contributions and highlights are presented in this work:(1)a multi-attribute two-layer network model is presented including the relationship of scholar cooperation behaviors and paper citation.(2)an influence factor R_a is defined as a“bridge”to connect paper’s citations to scholars and thus coupling the citations network and the collaborations network to evaluate the scholars’influence in global perspective.(3)a scholar influence representation model(SIRM)is constructed by integrating three aspects information to predict and identify high influential scholars.This work integrates many factors from the perspective of a two-layer network,and uses the modeling method of nonlinear analysis to reflect the effect of cooperation and citation on scholars’influence,which solves the problem of one sidedness caused by considering only a single network.The experimental results show that SIRM model can identify and predict highly influential scholars more accurately than other methods.Therefore,the high-impact scholar identification method proposed in this work can more accurately and comprehensively evaluate the influence of scholars,and provide a reference for identifying important scholars,guiding scholars to carry out scientific research and resource allocation.