基于神经网络的电子邮件分类与过滤

现在电子邮件的应用非常广泛，已经成为人们生活中一种重要的通讯手段，但各种各样的垃圾邮件也是令我们十分困扰的问题，给出了一种电子邮件的分类过滤方法。电子邮件作为一种半结构化的文档，电子邮件信息包含了固定的语法部分和一定长度的可变文本部分，同时处理这两部分以得到更高的准确度。首先对邮件进行文本处理，得到特征向量；然后使用基于神经网络的方法对邮件进行分类过滤得到邮件分类器；最后通过实验验证分类器的有效性。

E-mail classification and filtering based on neural network

Now electronic mail(E-mail) is widelyused,Which has becomeone of the most important forms ofcommunication available,at the same time junk mail is quickly becoming a serious problem.An E-mail classifying and filtering method is given.Modeled as semi-structured documents,E-mail messagesconsist of a setof fields with predefined semantics anda number ofvariable length free-text fields.By dealing with both of them,higher accuracy is obtained.First,E-mail is dealt with and features vector are got.Then method based on neural network is used to classify and filter personal E-mails.The result showed that the classifier can perform filtering with reasonable accuracy.