【摘要】 提出一种基于多特征模糊关联的垃圾邮件过滤方法.该方法分为预处理和实时处理两个阶段,在预处理阶段,分析训练样本集,提取邮件的发送源特征和文本特征的典型特征值集合,计算典型特征值与合法邮件类、垃圾邮件类之间的模糊关联度.在实时处理阶段,根据待分类邮件所包含的特征值,计算邮件的类支持度,然后利用Dempster-Shafer证据理论实现多个特征的分类融合与判决.实验结果表明,此方法能有效提高垃圾邮件过滤的查全率和查准率.更多还原

【Abstract】 This paper proposed a spam filtering method based on fuzzy relevancy of multiple features, it consists of two stages: preprocessing and real-time processing. In the preprocessing stage, it analyzes training samples set, extracts typical terms of email transmission source and text features, calculates fuzzy relevancy between typical terms and email categories. In the real-time processing state, it calculates support degree for categories according to the term values in email, then uses Dempster-Shafer evidence theory to fuse the information provided by multiple features and make decision. The experiment showed that the method can effectively improve the recall rate and precision rate of spam filtering.更多还原