Research On Role Recognition Of Research Team Members Based On Complex Network Analysis

Under the background of increasingly diversified knowledge,the scale of scientific research has been continuously expanded,and methods have been continuously updated.The research model also changed from early individual research to collaborative research.Under this trend,researchers in various fields spontaneously formed research teams with talent integration and knowledge integration.As the most important scientific production mode,the quality and development of scientific research team has become the focus of attention of many researchers.How to improve the quality of scientific research and how to manage a team gradually become the focus.In recent years,domestic and foreign scholars have done some research on the research team’s mining and scientific management,and put forward some more complete team evaluation index system,but have not given a better way to identify the roles of team members.In the past research,leaders’ identification mostly existed as one aspect of team management and team evaluation.Now they are gradually being studied separately.Scholars usually choose qualitative analysis of the impact of central indicators on the position of team members,or only consider a few social network centrality indicators,which are not comprehensive and rigorous.Therefore,the paper chooses a more comprehensive and reasonable centrality indicator to vectorize the role of team members,and then uses a sorting algorithm to build a model to identify academic leaders and core members in the team.In the recognition process,the paper chooses to use the classical RankSVM sorting algorithm for machine learning.It automatically combines team members,trains the sorting model,and then uses the model to test the team members.Finally,the importance of each member in the team is obtained.After sorting the importance values in descending order,the accuracy of the sequence is used as a measure of the quality of the algorithm.After a series of experimental analysis,the proposed method based on social network analysis using the RankSVM algorithm for character recognition is more reasonable and effective than using a single indicator to identify the role.The average accuracy of the final recognition can reach 73.3%.