The Design and Construction of a Relevent-Feedback-Based Search Agent for Enhancing Search Performance

• Main Authors: Yen Jia-Jane, 嚴嘉錚
• Other Authors: Jih-Shih Hsu
• Format: Others
• Language: zh-TW
• Published: 1998
• Online Access: http://ndltd.ncl.edu.tw/handle/17449043188670032741

碩士 === 國立雲林科技大學 === 資訊管理研究所 === 86 === Since the improper design of the user interface and the searching procedure, most search engines often make users, especially those without proper mental model about the search topics, have difficulty to decide which keyword would best address their search topics or to apply which keyword search further. There are similar findings in relevant papers about the problem of forming searching keyword concept, too. In addition, the high recall and low precision feature of ordinary search engines cause the inefficient search performance and make users waste a lot of time in reading irrelevant web pages. This thesis intends to design a special search agent based on the generic algorithm and machine learning process. This agent will automatically generate relevant keywords for recommendation and allow users to pick the appropriate keywords from the recommended keywords to search further.