Adaptive News Categorization

• Main Author: 王稔志
• Other Authors: Jason S. Chang
• Format: Others
• Language: zh-TW
• Published: 2001
• Online Access: http://ndltd.ncl.edu.tw/handle/83798425781857189990

碩士 === 國立清華大學 === 資訊工程學系 === 89 === This paper presents an unsupervised learning algorithm for automatic text categorization. We introduce stastistical constraints to filter features from description of each category. Experiment indicated that the constraints were very effective in obtaining the features that are relevant to the categories. Also , we propose a statistical relevance feedback algorithm to add more features to each category. The categorization accuracy exceeding 80 percent indicates that relevance feedback is effective for text categorization.