本文针对垃圾邮件过滤问题,结合中文自身的特点,把广泛适用于英文文本和邮件分类的朴素贝叶斯过滤方法应用在垃圾邮件网关邮件过滤层;把信息增益修剪方法经过改进作为中文特征选择方法,应用在数据管理层;从而极大提高了垃圾邮件的过滤精度。 关键字:朴素贝叶斯;信息增益; 特征提取; Abstract: To filter spam combined with Chinese we apply Naïve Bayes Algorithm to e-mail filtering layer, Information Gain as the basic feature-pruning method to Data Management Layer. In fact these two methods greatly improve the precision of spam filtering. Keywords: Naïve Bayes; Information Gain.