Design and Implementation of Personalized E-mail Classifier

The authors discuss the design and implementation of a personalized E-mail classifier based on Bayesian method, propose the architecture of the system and the algorithm of feature selection, and put forward a method to calculate posterior probability of the classification of new mails. According to the result of experiments, combining odds ratio feature sub-set selection method with naive Bayesian method has wonderful classification precision. When a new mail arrives, the system can classify it fast.