网络新闻分类系统及个性化新闻网站的研究与应用.pdfVIP

The information in the Internet is growing fastly along with the rapidly advancing of Internet. But facing the mess of information got from the Internet, the traditional way which deal with the information by human could not meet the actually demand. In addition, The website equip the character to facilitate the user will has more profit and is the key manager cared. Base on the need above, this paper introduce the web-mining technique to implement the functions below: information fast processing and website equipping the individuation character. Firstly, the theory and technique relative to the Chinese text auto categorization was deeply researched in this thesis. It includes webpage information extraction technique, text word-cut technique, text vectoring technique and character extraction technique. It designs the webpage information extraction arithmetic, selects max-match word-cut arithmetic, TF- IDF text vectoring arithmetic, relative information arithmetic, KNN text-categorization arithmetic, and improves the relative information arithmetic in this thesis. Secondly, it deeply researchs the Lucene text retrieve system and implements the Chinese word-cut function for Lucene in this thesis. Lucene system could add the text retrieve function to the text categorization system. Basing on the selected arithmetic, the text auto categorization system which could auto classify the html and txt document was developed by using the Java and OOP(design pattern etc)combining Lucene. The text- categorization system provides extenable interface for further improing and extending. After developing system it use international test criterion (recall, precision, and F1 test ) to test the System and detailly analyses the result of test in this thesis. At last, it researchs the user info