基于协同过滤的垃圾邮件过滤系统

Spam Filtering System Based on Collaborative Filtering

针对当前垃圾邮件过滤技术中邮件性质评价难及邮件附件判断难的问题,提出协同过滤模型,设计一个多层次垃圾邮件过滤系统。该系统针对垃圾邮件一般群发给多人的特点,提取用户的操作和阅读速度进行反馈,利用他人的反馈结果进行协同过滤。实验和分析结果表明,协同过滤技术有效地提高了垃圾邮件过滤的召回率。

This paper proposes a collaborative-based spam filtering scheme to overcome the quality evaluation problem existed in current spam filtering techniques. The scheme can be used to solve the problems with current techniques in classifying Email attachments. Because spam senders often resort to mass mailing, the system improves filtering accuracy by considering the factors that include user actions and reading speed as well as feedbacks from others on the same mailing list. The collaborative-based spam filtering is implemented as part of a hybrid spam filtering system. Experimental results show that the scheme significantly reduces the recall rate.