The Hong Kong University of Science and Technology
Department of Computer Science and Engineering


PhD Thesis Defence


"Learning-based Web Query Understanding"

By

Mr. Dou Shen


Abstract

Web search is attracting more and more attention with the exponential
increase in data becoming available on the Web. Typically, a Web user
submits a short Web query consisting of a few words to a Web search system
and expects the system to return the desired information. How to
understand Web queries precisely and thoroughly lies at the heart of Web
search.

In this thesis, we study the Web query understanding problem from several
viewpoints. The first one is through query topic classification, which
aims at interpreting queries in terms of predefined categories based on
the queries' topics. We propose three solutions for different scenarios in
query topic classification. The second way of understanding queries is to
detect the type of a Web query, such as whether it requires locality
information or information about a person, or the like. In this thesis, we
focus on personal name detection in Web queries. The third way is to
construct query hierarchies from query logs. These hierarchies reflect the
generalization/specification relationship among Web queries. Finally, we
study the contribution of query logs to Web-page classification, which can
in turn help understand Web queries. Experimental results on some
real-world data sets validate the effectiveness of our proposed methods.


Date:			Friday, 8 June 2007

Time:			10:00 a.m-12:00noon

Venue:			Room 3501
			Lifts 25-26

Chairman:		Prof. Shu Ming Ng (ISMT)

Committee Members:	Prof. Qiang Yang (Supervisor)
			Prof. Lei Chen
			Prof. Dik-Lun Lee
			Prof. Weichuan Yu (ECE)
			Prof. Lee-Feng Chien (National Taiwan Univ.,Taiwan)


**** ALL are Welcome ****