基于词序列核的垃圾邮件过滤

基于词序列核的垃圾邮件过滤 The Spam Filtering Based on Word-Sequence Kernel 陈葡 CHEN Pu 1 first-author 谢晓尧 XIE Xiaoyao 1 徐洋 XU Yang 1 贵州师范大学贵州省信息与计算科学重点实验室 贵州师范大学贵州省信息与计算科学重点实验室 Key Laboratory of Information and Computing Science of Guizhou Province,Guizhou Normal University Key Laboratory of Information and Computing Science of Guizhou Province,Guizhou Normal University 针对传统的基于词频特征向量核方法的垃圾邮件过滤算法中忽略词汇间的序列信息而导致信息损失影响过滤精度的问题,本文将词序列核与SVM（support vector machines）算法结合,对垃圾邮件进行过滤,相应的实验表明,该方法提高召回率、正确率和精确率,从而提高了过滤精度. In spam filter algorithm,to counter the situation where the traditional kernel methof basing on word frequency feature vectors the sequence information between vocabularies,which leads to the information loss and eventually influences the filtering precision,the article combines word nuclear with SVM algorithm,brings them into spam filtering and carries on the relevant experiment whose results show that the method is able to greatly improve the filtering precision. 词序列核 垃圾短信过滤 核方法 SVM word-sequence kernel spam filter kernel algorithm SVM 10.14188/j.1671-8836.2011.05.009 TP393.098 贵州省工业攻关项目（黔科合GY字[2008]3009）;贵州省科学技术基金（黔科合J字[2011]2213）资助项目;贵州师范大学资助博士科研项目 2011-05-01 2021-04-01 5 57