Research On Relatively Important Nodes Mining And Its Application In Criminal Networks

Relatively important node mining in complex networks has important research significance and a wide range of applications,such as finding remaining criminals through known criminals in criminal networks,and discovering possible pathogenic genes through known pathogenic genes in protein networks.This thesis mainly studies the mining algorithm of relatively important nodes,and how to improve the efficiency and accuracy of mining,and applies the algorithm to suspect mining in criminal networks.The main work contents are as follows:First,this thesis proposes a relatively important node mining method(RWR*)based on random walk with restart.This method mainly uses the idea of restarting the random walk to transmit the importance of the node.The node has a certain probability of jumping back to the starting point and restarting the walk at each step of the random walk.The initial value of the random walk is redesigned to improve the Probabilistic transition matrix for random walks.Experiments show that the RWR* algorithm proposed in this thesis has certain effectiveness.Secondly,with the development of the times,the scale of the network continues to expand.In order to improve the efficiency and accuracy of relatively important node mining,this thesis also proposes a local community discovery algorithm(SNC)based on the coincidence degree of known important nodes and second-order neighbors.The algorithm uses the known important nodes in the relatively important node mining model as specific nodes,and performs local community discovery based on the second-order neighbor coincidence degree.Experiments show that the SNC algorithm in this thesis has certain effectiveness,especially in some networks with few clique structures or obvious core nodes,the algorithm can show obvious advantages.Finally,apply the algorithm to criminal networks.In the more classic 911 criminal network,the RWR* algorithm is directly used for suspect mining.In the ICM criminal network,the network is first constructed,and the SNC algorithm is used to find out the local community to which the known criminal belongs,and then the RWR* algorithm is used in the local community to conduct suspect mining.Application experiments show that the RWR* algorithm can accurately use known criminals to mine unknown suspects on the 911 criminal network data.On the ICM criminal network data,the SNC algorithm can accurately discover the suspect community based on a small number of known criminals,which can effectively reduce the scope of mining and improve the accuracy of suspect mining.