一种相对重要节点挖掘方法及其在犯罪网络中的应用  

A Relatively Important Node Mining Method and Its Application in Criminal Networks

作　　者：罗正国 郁湧[1,2] 于倩 王莹港[1] 周彪 高涛 

机构地区：[1]云南大学软件学院,云南 昆明 [2]云南省软件工程重点实验室,云南 昆明 [3]云南经济管理学院教育学院,云南 昆明

出　　处：《计算机科学与应用》2021年第12期3028-3037,共10页Computer Science and Application

摘　　要：网络科学中相对重要节点的挖掘具有重要的应用价值,例如,通过已知犯罪分子查找其他犯罪分子。为此,文中提出了一种基于带重启随机游走的相对重要节点挖掘方法,并结合犯罪网络的两个特征来把该方法应用于犯罪网络的挖掘:根据犯罪分子会隐藏自己的特点对随机游走的初值进行设计;根据犯罪分子之间的通信人物很有可能就是罪犯的特点对随机游走的概率转移矩阵进行改进。最后将提出的方法同4个经典的相对重要节点挖掘指标进行对比,同时通过实际的犯罪网络来进行实验和分析,结果表明,文中的算法比经典的方法更能准确预测出犯罪分子。The mining of relatively important nodes in network science has important application value, for example, searching for other criminals through known criminals. For this reason, this paper proposes a relatively important node mining method based on random walk with restart, and combines the two characteristics of criminal networks to apply this method to criminal network mining: Design the initial value of the random walk according to the characteristic of criminals hiding themselves;Improve the probability transition matrix of random walk based on the characteristic of the communication between criminals who are likely to be criminals. Finally, the proposed method is compared with four classic relatively important node mining indicators. At the same time, experiments and analysis are carried out through the actual criminal network. The results show that the algorithm in this paper can predict criminals more accurately than the classic method.

关 键 词：复杂网络 相对重要性 节点挖掘 随机游走 犯罪网络 

分 类 号：D92[政治法律—法学]