基于最优属性约简的中文邮件过滤系统  

Filtering Spam System Based on the Dependent Rough Set Most Nature Reduction of Attributes

作　　者：李六杏[1] 区凤霞[2] LI Liu-xing,OU Feng-xia （1.Anhui administration institute,Hefei 230031,China;2,Anhui Chinese medicine institute,Hefei 230031,China）

机构地区：[1]安徽行政学院,安徽合肥230031 [2]安徽中医学院,安徽合肥230031

出　　处：《电脑知识与技术》2007年第3期1240-1241,共2页Computer Knowledge and Technology

摘　　要：对朴素贝叶斯理论作为中文邮件过滤技术进行了分析改进，邮件预处理后，对其进行分词处理，利用基于依赖性的粗糙集最优属性约简方法来对邮件集进行特征维数压缩。条件属性的数目大幅减少，提高了分类的效率。This article has analyised the Bayesian theory and proposed a way of improving its filterifig technique against Chinese mails,After pre-handling the mails, it will deal with them by phrases and then compress the characteristic dimension of the mail collection by using the reduction method of the best attribute of the dependent rough set. The number of the condition attributes reduces sharply, which thereby promotes the efficiency of classifying the mails.

关 键 词：垃圾邮件 中文分词 粗糙集 属性约简 贝叶斯分类器 

分 类 号：TP393[自动化与计算机技术—计算机应用技术;自动化与计算机技术—计算机科学与技术]