Research On The Discovery Of Criminal Gangs Based On Social Networks

Social network community discovery is an important branch of complex network research and has a wide range of applications.The discovery of criminal gangs is one of the key issues in the investigation and handling of public security organs.However,the methods of discovering criminal gangs based on social network analysis or attribute clustering analysis alone cannot effectively discover and mine unknown criminal gangs in big data sets.The performance of criminal gangs in social networks must be the community of social networks.Therefore,this paper first divides the social network data into communities and narrows the scope of gang identification.Based on the additional information such as structural similarity and attributes of criminal suspects in social networks,this paper studies the methods of discovering criminal gangs based on social networks.The main work of the thesis includes:Firstly,according to the structural characteristics of social networks,criminal gang characteristics and gang analysis methods,the idea of using social network modeling analysis to discover criminal gangs is proposed.Secondly,when the Louvain algorithm is used to divide the real social network,there is a redundant computing problem that generates modularity gain when the node moves.An improved Louvain algorithm based on preprocessing is proposed.The node degree storage adjacency list is calculated in advance.The nodes with different degrees are classified and judged.Improve the efficiency of large-scale network partitioning without changing the effect of community partitioning.Finally,based on the advantages of the improved Louvain algorithm and the structural characteristics of large-scale social network data,a gang discovery algorithm model combining topological similarity is proposed.Through the community discovery and social network node(entity)attribute information,a gang discovery algorithm model combining attribute information is proposed.Experiments have shown that the improved Louvain algorithm can improve the efficiency of social network community partitioning.The gang discovery algorithm model combined with topological similarity can effectively identify criminal gangs such as telecom fraud according to the cosine similarity.The gang discovery algorithm model combined with attribute information can further narrow down and determine potential criminal gangs based on node attribute information.These tasks can provide effective theoretical and technical support for public security organs to discover suspicious criminal gangs.