Research On Specific Group Discovery Technology In Social Networks

With the rapid development of Internet information technology,there have been groups on social networks that spread harmful information,commit crimes and organize terrorist activities.These groups often use social networks to connect,which often has serious negative effects.As an important part of social network research,group discovery is not only important for understanding the characteristics of group structure,revealing hidden information of the group,conducting group guidance and control,but also has a positive role in eliminating the negative effects of group activities.This article has conducted the following three aspects of research on how to quickly and accurately discover specific groups in social networks.Not only provides new ideas for group discovery,but also provides support for public security organs to crack down on and control online group crimes.(1)Aiming at the discovery of specific target groups in social networks,a local extension group discovery algorithm based on specific nodes(SNLEGDA)is proposed.The algorithm first uses the associated network formed by specific nodes to conduct a comprehensive centrality analysis of the seed nodes,then expands the group based on the local modularity,and finally merges the groups,so as to quickly find the group where the specific node is located.Experiments on the Zachary network data set prove the superiority of the algorithm.Experiments on the Football network,Dolphins network,and MLM group network data set prove the effectiveness of the algorithm.(2)Aiming at the problem of low accuracy of the group discovery algorithm based on network structure,a group extension and correction algorithm based on attribute similarity(ASGECA)is proposed.The algorithm first analyzes the attribute characteristics,then designs a dynamic weight similarity measurement method based on node attributes,and finally expands and revises the group,in order to find the group closer to the real group.Experiments on the MLM group network data set prove that the algorithm can improve the accuracy of group discovery results.(3)To apply the SNLEGDA and the ASGECA to actual combat,design and implement a specific group discovery prototype system(SGDPS).Through Django + MySQL + Echarts development mode,Elasticsearch,Python combined with HTML,CSS,JS and other technologies to complete the system development.The functions of the two algorithms in the research process are implemented in the system,including group preview,group relationship association,seed node analysis,group discovery,group expansion and correction.Finally,the prototype system is tested.The test results show that the prototype system can accurately find the group where the specific node is located,which proves the availability of the system.