基于兴趣相似度的协作式垃圾邮件过滤方案  

A Collaborative Spam Filter Proposal Based on Interest Similarity

作　　者：潘蓝兰[1] 薛开平[1] 洪佩琳[1] 

机构地区：[1]中国科学技术大学电子工程与信息科学系信息网络实验室,安徽合肥230027

出　　处：《通信技术》2010年第2期122-124,共3页Communications Technology

基　　金：国家自然科学基金(60772033);安徽省自然基金(090412048);安徽省优秀青年人才基金重点项目(2009SQRZ004ZD)资助

摘　　要：由于电子邮件在互联网上的广泛应用,出现了垃圾邮件的泛滥和占用网络带宽,垃圾邮件给用户正常收发邮件造成了严重困扰。提出了一种新的基于兴趣相似度的协作式邮件过滤方案,使得兴趣相似的用户迅速聚合成信息共享的团体,每个用户在团体内选择出最合适的过滤规则集进行有效的共享。实验表明,该方案在保持一定的过滤精度的同时有效降低了误判率。Due to the wide application of e-mail on the Internet, the spam is rampant and occupies the network bandwidty. Spam is a big trouble when people open their mailbox. In this paper, a new collaborative spam filtering proposal based on interest similarity is proposed. Similar users are quickly clustered into the different information-sharing communities. Each user could select the most suitable filtering rule set from the community for sharing. The results of experiment show that the proposal could lower the false-positive effectively while maintain a certain level of spam-filtering accuracy.

关 键 词：垃圾邮件 共享 兴趣相似度 协作式过滤 

分 类 号：TP309[自动化与计算机技术—计算机系统结构]