个性化E-mail分类器的设计与实现

Design and Implementation of Personalized E-mail Classifier

详细讨论了一个基于朴素贝叶斯方法的个性化E-mail分类器的设计,给出了系统体系结构和特征提取算法,试验了一种对新邮件计算所属类别后验概率的方法.试验结果表明,结合Odds Ratio特征子集提取算法和仆素贝叶斯方法对邮件进行分类具有较好的分类精度.应用朴素贝叶斯方法在新邮件到达的同时对其进行分类,具有较好的分类速度.

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.