一个改进的中文文本过滤系统的设计和实现.pdfVIP

一个改进的中文文本过滤系统的设计与实现 李东林迟呈英战学刚 鞍山科技大学计算机科学与工程学院，鞍山 114044 摘要： 中文文本过滤系统的研究与设计是当务之急．针对传统文本过滤仅仅把文档相似度同设定阈 值相比较判定相关与否的判断方法在过滤性能方面的不足，本文设计了一个改进的文本过滤系统．系 统对流入文档进行两重过滤。通过引入用户主题不相关配置文件和相应的不相关阈值，进一步的对传 统文本过滤返回的结果文档集进行不相关性检测，剔除与用户主题不相关的配置文件相关度大于不相 关闽值的文档，减少不相关文档的返回。实验证明．该系统提高了文本过滤性能． 关键词：文本过滤；向量空间模型；用户模板：文本特征抽取 and ofan DesignImplementationImproved Document FilteringSystem Li Chi Zhan DonglinChengyinXuegang and InstituteofScienceand 114044 Schoolof Science Technology,Anshan Computer Engineering．Anshan and of informationis Document Abstract：DesignChinese filteringimperative．Traditional analysis the between documentanduser with threshold， comparessimilarityincoming profilesystem’S FilteringSystem which are thanthethresholdtotheuser．This andthen documentssimilarities higlIcr designed provides paper an Document for includestwo system filtering improved FilteringSystemimprovingperformance．The user’snon-relevantandnon·relevant the threshold，the process．Byimporting profile systemcomputes result fromthefirsttraditional betweenuser’Snon·relevantandthe documents filtering similarity profile thedocumentssimilaritiesare thanthenon·relevantthreshold． then which rcjects