计算机工程与应用 ›› 2017, Vol. 53 ›› Issue (6): 96-100.DOI: 10.3778/j.issn.1002-8331.1508-0118

• 大数据与云计算 • 上一篇    下一篇

基于网络关系的微博水军集团发现方法

叶施仁,叶仁明,朱明峰   

  1. 常州大学 信息科学与工程学院,江苏 常州 213164
  • 出版日期:2017-03-15 发布日期:2017-05-11

Method to find spammer group for Weibo based on network relationship

YE Shiren, YE Renming, ZHU Mingfeng   

  1. School of Information Science & Engineering, Changzhou University, Changzhou, Jiangsu 213164, China
  • Online:2017-03-15 Published:2017-05-11

摘要: 由于目前水军的高伪装性,经典的水军识别算法变得不再有效。与真实用户相同,水军用户之间也会形成一定的网络结构,提出了一种基于网络关系的方法来发现水军集团,首先以一个典型的水军账号作为种子,逐层扩展粉丝关系,优先搜索出现次数频繁的用户,从而获得一个包含大量水军账号的集合,按照水军用户之间关系的高度聚集性以及与真实用户之间关系稀疏性的特点,用Fast Unfolding算法进行社区检测。实验结果表明,该方法能够很好地发现水军集团。

关键词: 水军集团, 网络关系, 社区检测

Abstract: Due to high camouflage of current spammer, classic spammer discovered algorithms become no longer valid. Same with real users, spammers can also form a network structure. This paper proposes an algorithm based on the network relationship to find spammer group. The first step is to find a typical spammer account as seed. Then it extends fans’ relationship step by step and searches account which occurs frequently to get a collection which contains a large amount of spammer accounts, according to the clustering of the relationship between spammers and the characteristics of the sparse nature between spammer and real users, using Fast Unfolding algorithm to community detection. Through experimental analysis, the method can well find spammer group.

Key words: spammer group, network relationship, community detection