E-Mail Filtering Based on Analysis of Structural Features and Text Classification

Concerning the requirement of e-mail filtering to improve the efficiency and accuracy in e-mail mining, topic detection, and many other specific applications, learnt from traditional spam filtering methods, an approach based on feature analysis and text classification is proposed. Utilizing some structural features which are very likely to identify an irrelevant e-mail, such as group sending, embedded pictures, and so on, feature analysis filtering makes up the disadvantage of spending too much in text classification. An idea of identifying the category of a group-sent mail by the presence of personal names is proposed and the method of e-mail filtering based on URL blacklist is improved. Considering the different contribution of subject and body text to the category, the algorithm of Naive Bayesian e-mail classification is improved. The experimental results show that the method is reasonable and effective.