Research On Scientific Research Team Discovery Based On Complex Network

With the advancement of science and technology and the promotion of globalization,the historical period of scientific research in which scientific research is relatively independent and lack of scientific research team has become the past.Scientific research cooperation has become the need to overcome scientific research difficulties and promote scientific and technological progress.Scientific research cooperation in different disciplines,different institutions,different regions and even different countries is very common.Scientific research cooperation is a kind of labor form in which scientific researchers cooperate with each other according to the plan in order to complete a certain scientific research task.Scientific research team is a scientific research working group with similar scientific research goals and objectives formed in the process of scientific research cooperation,which takes scientific and technological research as the content.It is of great significance for scientific research institutions to find out the scientific research team in a certain field and dynamically track the evolution trend of the scientific research team,so as to provide more direct changes in discipline construction.First of all,the thesis discusses and analyzes the existing community discovery algorithm model and the representation method of complex network.With the emergence of large-scale network,the traditional research and analysis of social network based on network topology has certain limitations.Compared with the traditional research and analysis method of social network,the network representation learning algorithm based on deep learning can better represent the complex network.Therefore,this thesis proposes a community discovery algorithm based on network representation learning.The algorithm gets the path of complex network by network walking and expresses the network as vector by using skip gram model.Finally,K-means algorithm is used to cluster the node vectors of the network to discover the community.Experiments are carried out on 3 classic community discovery datasets and 2 scientific research co-authored network datasets.The effectiveness of the community discovery algorithm is verified by comparing with other community discovery algorithmsFinally,the thesis uses 20-year literature in the field of Library and information science as data,and takes four years as a stage to study the evolution of the scale of scientific research team and the change of the cooperation mode of scientific research team,and analyzes the change of scholars in scientific research team through four aspects: the characteristics,cooperation degree,the amount of scholars’ publications and the degree of scholars’ centrality.The thesis summarizes the evolution of scientific research team and scholars in the field of Library and information in the past 20 years,and finally takes the library and information teachers of Xi Dian University as an example to analyze the visualization and evolution of scientific research team.