一个初审/复审协作式反垃圾邮件系统  

A Collaborative Anti-spam System Based on First-Second Filtering

作　　者：王兴伟[1] 张登科[2] 温占考[2] 易秀双[2] 

机构地区：[1]东北大学信息科学与工程学院 [2]东北大学网络中心,辽宁沈阳110004

出　　处：《厦门大学学报（自然科学版）》2007年第A02期88-91,共4页Journal of Xiamen University：Natural Science

基　　金：国家242信息安全计划(2005C59)资助

摘　　要：提出了一个反垃圾邮件系统的三层模型,并使用初审/复审流程建立了一种协作式反垃圾邮件系统功能框架.在该功能框架下,初审使用基于规则匹配的邮件过滤技术,快速处理海量邮件,将邮件分成正常、垃圾和疑似垃圾邮件三类;复审使用内容分析技术对疑似垃圾邮件进行甄别,同时使用基于反馈学习的自适应技术更新初审规则.基于该功能框架,设计并实现了一个反垃圾邮件系统,应用在邮件系统中进行了实测.实验结果表明,该系统可以有效地处理垃圾邮件,系统的漏报率<3.7%,误报率<0.6%,性能高于公开的反垃圾邮件产品性能.A three-level anti-spam model and a novel anti-spam function framework based on first-second filtering are introduced in this paper. In the first pass, rule-based filtering method is used to process the mass mails into three classes： ham and spam and suspected spam quickly. In the second pass, content analysis method is used to discriminate the suspected spam further and the adaptive method based on feedback learning is used to update the rules used in the first pass. An anti-spam system has been implemented based on the above model and flamework. Experiment results have shown that it is effective and efficient. Its miss rate is lower than 3.7% , false rate is lower than 0.6% ,and the performance is better than other anti-spam products.

关 键 词：垃圾邮件 初审 复审 规则 内容分析 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]