一种新的犯罪团伙挖掘算法  

A New Criminal Gang Mining Algorithm

作　　者：唐德权[1] 黄金贵 史伟奇[1] TANG Dequan;HUANG Jingui;SHI Weiqi(Department of Information Technology,Hunan Police Academy,Changsha 410138,China;College of Information Science and Engineering,Hunan Normal University,Changsha 410081,China)

机构地区：[1]湖南警察学院信息技术(网监)系,湖南长沙410138 [2]湖南师范大学信息科学与工程学院,湖南长沙410081

出　　处：《湖南科技大学学报（自然科学版）》2023年第2期80-87,共8页Journal of Hunan University of Science And Technology：Natural Science Edition

基　　金：湖南省教育科学“十四五”规划课题资助项目(XJK23BGD034);湖南警察学院高层次人才科研启动专项基金资助项目(2022KYQD03);国家自然科学基金资助项目(61471169)。

摘　　要：为了利用图模式挖掘犯罪情报网络中的核心团伙和核心人物,提高犯罪网络威胁预测和识别的效率,提出一种新的核心团伙挖掘算法(Core Gang Mining Algorithm,CGMA).对海量的犯罪情报网络数据集建立相应的无向简单图模型,通过改进图挖掘方式,构建候选核心团伙集的数据结构,并提出由k-团伙通过连接和扩展2种操作得到(k+1)-团伙,从各个不同的图数据中统计其频度,最后在模拟数据集和真实数据集上验证算法CGMA的准确性和时间复杂度.该算法避免了传统的图模式挖掘中的子图同构问题,同时也优于其他常用的犯罪团伙挖掘算法.试验结果表明:该算法能对犯罪核心团伙信息进行有效预测.In order to utilize graph patterns to mine core gangs and key characters in criminal intelligence networks and improve the efficiency of threat prediction and identification in criminal networks,a new core gang mining algorithm(CGMA)is proposed.Data set based on vast amounts of criminal intelligence network,the undirected simple graph model is established,by improving the mining method and building candidate core gangs set data structure.It obtains the k-gangs by join and extension of two(k+1)-gang from different graph data statistics of the frequency.Finally,the accuracy and time complexity of the algorithm CGMA are verified on the real data set,the algorithm avoids the problem of subgraph isomorphism in traditional graph pattern mining,which is better than other common mining algorithms of criminal gangs.Experimental results show that the algorithm can effectively predict the information of criminal core groups.

关 键 词：图模式 核心团伙 图挖掘 连接和扩展 子图同构 

分 类 号：TP311.2[自动化与计算机技术—计算机软件与理论]