基于用户关系挖掘和信誉评价的垃圾邮件识别算法  

A spam identification algorithm based on user relationship mining and reputation evaluation

作　　者：王巍[1] 苘大鹏[1] 玄世昌[1] 杨武[1] 邱文真 

机构地区：[1]哈尔滨工程大学信息安全研究中心,哈尔滨150001

出　　处：《高技术通讯》2012年第9期889-896,共8页Chinese High Technology Letters

基　　金：863计划（2007AA01Z473）和中央高校基本科研业务费专项资金（HEUCF100601）资助项目.

摘　　要：采用基于行为的垃圾邮件识别思想，提出了一种基于用户关系挖掘与信誉评价的垃圾邮件识别算法。首先在分析邮件用户群体历史通信关系的基础上，建立邮件用户关系描述模型。进行邮件信誉评价时，利用用户关系描述模型对潜在用户通信关系进行挖掘，并根据邮件接收用户和邮件指纹建立邮件通信路径集，根据路径集中的用户对该邮件对应指纹的历史评价结果得到其综合信誉值，以识别垃圾邮件。在收件人进行反馈时，实时更新用户和邮件指纹信誉值。实验结果表明，该算法识别垃圾邮件的准确率达到95％以上。在算法设计中考虑了实用性，可用于实际的垃圾邮件识别系统。According to behavior-based spam identification ideas, a spam identification algorithm based on user relation- ship mining and reputation evaluation is proposed. First, a mail user relationship description model is established on the basis of analyzing historical communication in the mail user group. When mail reputation is evaluated, the potential user communication relationship is mined using the model proposed, and the mail communication path col- lections are established according to mail receiving users and mail fingerprints. The comprehensive reputation value is calculated to identify spam according to historical evaluation of mail fingerprints given by users in the path collec- tions. When a mail receiver gives his feedback, the reputation of user and mail fingerprints is updated. The experi- mental results show that the proposed algorithm can identify spam correctly and effectively. The availability is con- sidered in the design of the algorithm, which makes it work in actual spam identification systems.

关 键 词：垃圾邮件 行为分析 用户关系挖掘 信誉评价 

分 类 号：TP393.098[自动化与计算机技术—计算机应用技术]