Key Technology Research On Content-Based Text Filtering

The 21^st century is a new epoch of information technology, and it is also a new epoch ofInternet. With the advances in communication technology and the popularization of Internet,the amount of information for users will become increasingly important, especially on theWorld Wide Web. Internet brings a chance for users to discover much more information thanever, but on the other hand users often feel confused facing such a disorder, huge informationspace which contains thousands upon thousands network links.Filtering is one method to help users to obtain the information that mostly fits theirneeds. The function of information filtering (IF) is to select relevant information or eliminateirrelevant information from dynamic information flow on the Internet according to certaincriteria and some approaches. Information filtering is the precondition to realizeindividualized information service. Different users have different information needs, soindividualized information service offer different services to meet their needs. This process isalso voluntary. The purpose of individualized information service is to offer users bestservices, meanwhile the users afford little. This thesis commences the basic concept of information filtering, discusses theprinciple and basic processing procedure of information filtering, summarizes theclassification of filtering systems, analyzes several classic information retrieval models andintroduces the methods about how to evaluate the performance of filtering systems. Then keytechniques in content-based text filtering are particularly discussed, that is how to segmentChinese words, how to extract suitable features from documents, and how to construct userprofile, etc. Finally, the thesis proposes a design scheme of content-based text filteringprototype system and describes the details of its implementation.