摘要： 如何利用邮件的正文与附件信息有效地实现其分类，是现在邮件处理领域一个重要的课题。该文从商业应用角度提出了一种基于图像信息度量与关键词的邮件智能过滤与分发方法，通过基于朴素贝叶斯分类器的邮件关键词信息处理，及附件图像信息的基于归一化PIM文本图像检测理论的分析，能够综合运用邮件正文、地址等文本信息与附件图像信息作为分类的评价参数，有效地实现了邮件的智能分类。

关键词: 多用途网际邮件扩充协议, Base64, 图像信息度量, 朴素贝叶斯

Abstract: It is an important technology to classify emails by using text and appendix information in the field of email processing. This paper presents an intelligent email classification system based on PIM and keywords. Keywords are used as the input features of native Bayes classifier to filter text of email. PIM is used as the statistical feature to classify images in appendixes. Experiments show that this system is efficient in intelligent emails classification.

Key words: multipurpose Internet mail extensions(MIME), Base64, picture information measurement(PIM), Native Bayes

中图分类号: 

• TP391