摘要： 随着电商的不断发展，如何发现刷单账户以维护市场秩序是亟待解决的问题。根据用户的购买记录构建用户关系网络图。根据电商刷单特性，提出通过计算用户之间的关联性来对抗伪装，并提出了一种密集子图可疑度量。根据关联性和密集子图可疑度量，实现了一种针对电商的刷单团伙识别方法。实验结果表明，在真实的天猫数据集上，所提出的方法能有效地捕捉刷单团体, 结果与FRAUDAR算法相近，有较好的识别效果。

中图分类号： TP391
文献标识码： A
DOI： 10.19358/j.issn.2096-5133.2021.03.013
引用格式： 郑赜一，王澎. 基于密集子图挖掘的刷单团伙识别方法[J].信息技术与网络安全，2021，40(3)：72-76.

Click farm identification method based on dense subgraph mining

Abstract： With rapid development of e-commerce, how to find out click farm to maintain market order has become an urgent problem. This paper constructs a graph based on the purchase records of users. With the featrues of click farm, it proposes to combat camouflage by calculating the correlation between users, and puts forward a subgraph anomaly degree metric. Finally，a suspicious subgraph searching algorithm for e-commerce to find out click farm is proposed.The results show that the proposed algorithm can find out the click farm effectively, and it is similar with FRAUDAR on the real dateset of Tmall.

Key words : graph mining；dense subgraph；click farm；fraud identification