邮件过滤与个性化分类模型的研究  

Research on Email Filtering and Personality Classification Model

作　　者：王丽侠[1] 

机构地区：[1]浙江师范大学信息科学与工程学院,浙江金华321004

出　　处：《微机发展》2005年第9期42-44,47,共4页Microcomputer Development

基　　金：浙江省教育厅科研项目资助(20030717)

摘　　要：研究了邮件过滤的主要方法,提出了将Agent技术、粗糙集和最小风险的Bayes分类方法结合的邮件过滤及个性化分类模型。该模型首先利用粗糙集方法对邮件样本向量空间进行约简,然后利用已知样本对最小风险的Bayes分类器进行训练,得到具有智能分类功能的邮件分类器,利用该分类器过滤掉用户不感兴趣的邮件,并利用Agent学习用户的个性化知识,最后利用学习的知识将用户感兴趣邮件进行再分类。Email filtering methods were discussed. And a model was put forward which integrated agent technology, rough set and minimum risk naive Bayes approach. At first, sample vectors were reduced by rough set approach, then minimum risk naive Bayes filter was trained by training sample and intelligent Email filter was built which can filter junk Email, at last the agent can classify user interesting Email by learning user individual knowledge.

关 键 词：邮件过滤 贝叶斯分类 机器学习 代理 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]