基于面向概念漂移的垃圾邮件CBR过滤方法  

SPAM FILTERING APPROACH FOR CONCEPT DRIFT USING CBR

作　　者：刘莹[1] 

机构地区：[1]武汉长江工商学院湖北武汉430065

出　　处：《计算机应用与软件》2013年第9期296-299,322,共5页Computer Applications and Software

摘　　要：针对文本流分类中的概念漂移问题,以垃圾邮件过滤为应用背景,提出一种能适应概念漂移的垃圾邮件基于案例推理CBR(Case-based Reasoning)过滤算法。算法采用CBR过滤垃圾邮件,研究CBR过程中的案例库管理技术,提出基于惩罚降噪和等价除冗的案例库修正算法,以适应概念漂移问题。在真实数据集上的实验验证了提出的案例修正算法获得的垃圾邮件过滤效率的提高,可以更好地解决垃圾邮件中的概念漂移问题。For the concept drift problem in text stream classification and taking spam filtering as the application background, we propose a spam filtering algorithm using CBR （case-based reasoning）, which can adapt to concept drift. The algorithm uses CBR to filter the junk mails. We study the case base management techniques in CBR process, and propose a case base modification algorithm to adapt to the concept drift problem, it is based on punishment-based noise reduction algorithm and performance-based equivalence redundancy reduction algorithm. The experiment on real data sets validates the improvement in spam filtering efficiency gained by the proposed case modification algorithm. It can better tackle the concept drift problem in spam.

关 键 词：文本挖掘 垃圾邮件过滤 概念漂移 案例推理 案例库管理 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]