基于分词的垃圾邮件过滤系统设计与分析

基于分词的垃圾邮件过滤系统设计与分析 Design and Analysis of Spam-Filtering System Based on Words Segmentation 唐朝晖 TANG Zhao-hui 1 2 first-author 傅建明 FU Jian-ming 1 2 杜南山 DU Nan-shan 1 2 武汉大学计算机学院 武汉大学计算机学院 School of Computer,Wuhan University School of Computer,Wuhan University 武汉大学计算机学院,湖北,武汉,430072 武汉大学计算机学院,湖北,武汉,430072 以字为单位处理中文邮件存在着很大困难,针对于此,笔者引进中文分词算法,设计并实现了基于分词的垃圾邮件过滤系统,并且阐述了与实现相关的3个关键算法:用于关键词匹配的多模式相似/精确匹配算法,用于中文邮件处理的中文分词算法,以及用于特征提取的N元特征提取算法.最后实验证明了该系统对中英文垃圾邮件过滤都有很高的性能.另外,文章第三部分还给出了基于分词的非垃圾邮件分类系统的设计与实现. It is difficult to process Chinese Emails by single word.In order to resolve this problem,a spam-filtering system based on words segmentation is designed and implemented.Meanwhile,several related algorithms when implemented are demonstrated: Multiple Exact/Approximate String Matching Algorithm for matching,Chinese Words Segmentation Algorithm for processing Chinese emails and N-Gram Feature Extraction Algorithm for feature extraction.Finally,the experiment validates the high performance of processing Chinese Emails as well as that of English Emails.In addition,a sortion system for NonSpam is designed and implemented in the third part of the paper 垃圾邮件 多模式相似/精确匹配 中文分词 N元特征提取 spam multiple exact/approximate string matching chinese words segmentation N-gram feature extraction TP393.098 国家自然科学基金资助项目（90104005,60373089）;湖北省科技攻关项目（2002AA101C44） 2005-01-01 2021-04-01 S2