Research on intelligent agent for mail server
QI Xing-min · Jisuanji gongcheng yu sheji · 2007
The influence of E-mail structure on mail classification is analyzed,and bi-layer classification method applied to intelligent agent for mail sever is proposed including these modules: Mail length classification,collection and pretreatment of E-mail,Chinese words segmentation,feature selection and the classification of E-mail.This Agent can not only support the ability of automatically filter spam,but also apply to E-government E-business for automatic classification and transmission of E-mail.First mails are classified ac-cording to different length,and then are trained by different Bays classifier for filtering spam.Experiments indicated that the bi-layer classification method is of high classification precision and efficiency.