A Search Agent with Website Models

• Main Authors: Yung Ting, 丁勇
• Other Authors: C.S. Ho
• Format: Others
• Language: zh-TW
• Published: 2000
• Online Access: http://ndltd.ncl.edu.tw/handle/03934431744550187368

碩士 === 國立臺灣科技大學 === 電子工程系 === 88 === In this thesis, we purpose the concept of website models to help a search agent locate the most important information for users, efficiently and effectively. A website model is defined to be a structure that captures statistics on static and dynamic website information, including webpage descriptions, webpage clusters, website clusters, long-term policy, short-term activities, information update frequency, and response time. The search agent is responsible for website modeling, which involves the concepts of ontology and ontology-supported user models to help expand a given query. We find that the modified prTFIDF classifier supported by Ontology can fast and accurately classify retrieved webpages and websites during the construction of website models. With this high-quality website model, a search agent can do better quality information search.