The Design And Implementation Of Real-Life Scholar Community Detection System

Industry-university-research cooperation has promoted the formation of innovation networks.However,in the process of promoting industry-university-research cooperation,it is difficult for enterprises to obtain information on academic teams in colleges and universities.Among them,the large number of scholars and the information asymmetry between enterprises and universities are the main causes of problems.Scholars' cooperation data on the Internet can be used to build academic networks and explore the composition of academic teams to provide information support for enterprises.Currently,there are network analysis software on the market that can divide the community structure,but its function can not fully meet the needs of enterprises to understand the academic team.This thesis designs and implements a real-life community detection system to help users understand the composition of academic teams and important scholars in the team.The system builds an academic cooperation network by collecting scholarly collaboration data published on the Internet,and then explores the academic team composition and its important nodes from the network,and uses visualization technology to present the results to users,thus providing users with a decision support tool.The main work of this thesis is as follows:(1)A learning community detection model based on network embedding is proposed.The author's topic model is used to train the abstract of the thesis to obtain the scholar's subject probability distribution.By improving the calculation process of the transition probability in the node2 vec random walk,the text information of the node is integrated into the feature sequence extraction process to obtain the scholar vector,and then the K-means algorithm is improved.The clustering center selection and clustering number selection are used for scholar clustering,and finally each cluster is the scholar community.(2)Identification of important scholars in the academic team.Taking into account the scholar's social distance factor,social location factor and the ability to carry information flow,by calculating the degree center value,near center value and median center value of each node in the community,the average weighting is the node's comprehensive centrality,which helps to find academics.An important scholar within the team.(3)Design and implement a real-life community detection system,including social data subsystem with data collection function,data pre-processing function and network construction function,and social application with community division function,core node detection function and community visualization function.Subsystem.The function test and case analysis of the system are carried out to verify the effectiveness and feasibility of the system.