Mining Navigation Behaviors for Term Suggestion of Search Engines

碩士 === 國立臺灣科技大學 === 電子工程系 === 90 === Query expansion is a common technology for the information retrieval system, such as search engines. Most previous approaches of query expansion are based on the text analysis in the corpus or retrieved documents. However, many issues such as segment...

Full description

Bibliographic Details
Main Authors: Wei-Tang Hung, 洪偉棠
Other Authors: Hahn-Ming Lee
Format: Others
Language:zh-TW
Published: 2002
Online Access:http://ndltd.ncl.edu.tw/handle/63775768148539784368
Description
Summary:碩士 === 國立臺灣科技大學 === 電子工程系 === 90 === Query expansion is a common technology for the information retrieval system, such as search engines. Most previous approaches of query expansion are based on the text analysis in the corpus or retrieved documents. However, many issues such as segmentation and feature selection must be handled and the performance might be influenced easily. In this thesis, we investigate the ways to apply access logs in the search engines to the term suggestion. We propose a co-clicked behavior based term suggestion, which is designed to suggest user-oriented terms by the collaborative method. We also develop a search engine prototype in order to demonstrate the results of term suggestion. This thesis focuses on avoiding the issues of text analysis, which is used by common approaches of query expansion. By analyzing the co-clicked behaviors in the access logs, it does not need to perform text analysis and provides some good characteristics, which the previous approaches of query expansion lacked, such as content independent, adaptive, and extensible. Besides, limitations of current search engines such as word mismatch and partial match problems can also be overcome. The experimental result shows that, using our suggestion results the precision can be improved by 13.67%. Compared with another Chinese term suggestion system OPENFIND, our term suggestion also has the better performance.