一种用于电子邮件分类的特征选择算法  

Feature Selection Algorithm in Email Classification

作　　者：王卫玲[1] 刘培玉[1] 刘克非[1] 

机构地区：[1]山东师范大学信息科学与工程学院,济南250014

出　　处：《计算机工程与应用》2006年第33期122-124,156,共4页Computer Engineering and Applications

摘　　要：邮件分类是当前研究的一个热点问题,而如何进行邮件特征选择,是邮件分类中的重要问题。在介绍几种常用的邮件分类的特征选择算法的同时,提出了将非搜索型算法FCBF与搜索型算法SFS结合的特征选择方法。实验验证了该方法的有效性和可行性,能够有效提高分类器的准确率。How to classify incoming E-mails and filter spam has attracted much attention.One fundamental yet important issue in E-mail classification is how to select the appropriate features.Based on the analysis and comparison of several typical feature selection methods for E-mail classification,a new method has been proposed,which combines both FCBF and sequential forward selection.Experimental result shows that the proposed method can improve the precision of E-mail classification.

关 键 词：邮件分类 垃圾邮件 特征选择 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]