目前最流行的邮件内容过滤技术工作在网络应用层,通过对邮件内容的分析来判别邮件的合法性,无法避免由于垃圾邮件的泛滥而造成的网络带宽资源的浪费。针对这种情况,论文提出一种基于网络会话层的垃圾邮件行为识别方法。该方法运用决策树算法,对邮件发送过程中的网络会话层数据进行挖掘,发现垃圾邮件的行为规律,在垃圾邮件的内容数据发送前就对其实施过滤,有效地解决了垃圾邮件占用网络带宽的问题,是对当前各种垃圾邮件过滤技术的一个有益的补充。 At present, the most popular e-mail content filtering technology works at the network application layer. By analyzing the e-mail content to determine the legitimacy of e-mail, it is inevitable to waste network bandwidth resources caused by the flooding of spam. In response to this situation, the paper proposes a method of spam behavior identification based on web session layer. This method uses the decision tree algorithm to mine the network session data in the mail sending process and find out the rules of spam behavior. It filters the spam data before it is sent, which effectively solves the problem of occupying the network bandwidth The issue is a useful complement to the current spam filtering technologies.