基于贝叶斯最小风险分类的邮件过滤系统

E-mail Filter System Based on Bayesian Smallest Risk Classification

针对互联网上垃圾邮件给用户带来种种困扰的问题,本文提出了一种基于贝叶斯最小风险分类方法的邮件过滤系统。本方法通过设置损失代价函数,在过滤大部分垃圾邮件的同时,保证了将合法邮件保留,避免了将有用邮件误分类为垃圾邮件时,给用户带来的损失。实验结果表明,本文提出的垃圾邮件过滤系统效果较好。

To solve the problem that the rubbish E-mail brings so much trouble to the internet user, the paper proposes a E-mail filter system based on the Bayesian Smallest Risk Classification Method. The method not only removes most rubbish E-mails, but also saves every natural ones by setting the cost function decision table. So the proposed filter system avoids the losing that suffers from the error classification of useful E-mail to rubbish ones. The experiment results show that our E-mail filter system is more efficient.