Abstract: Content-based filtering is one of the mainstream technologies used so far.In view of which the essence of spam filter is the classify problem，this paper aims to provide a kind of filter method against spam based on SVM，and tries to adopt SMO algorithm to categorize the spam.Through the experiment，SMO algorithm can perform better，and the classification time of SVM classifier is reduced greatly.

Key words: spam, Support Vector Machine（SVM）, Sequential Minimal Optimization（SMO）, classification time

摘要： 基于邮件内容的过滤是当前解决垃圾邮件问题的主流技术之一。针对垃圾邮件过滤本质是分类问题，提出了一种基于支持向量机对垃圾邮件过滤的方法，并且将SMO分类算法结合到垃圾邮件分类中。通过实验，SMO算法能够取得较好的分类效果，缩短了支持向量机分类器的分类时间。

关键词: 垃圾邮件, 支持向量机, 序列最小优化算法, 分类时间