BYU

Abstract by Joseph Bills

Personal Infomation


Presenter's Name

Joseph Bills

Degree Level

Doctorate

Co-Authors

Dennis Ng

Abstract Infomation


Department

Computer Science

Faculty Advisor

Dennis Ng

Title

An Information Efficient Query Suggestion Approach Based on Implicit Structure of the Corresponding Corpus

Abstract

Diverse people use all sorts of search engines daily to retrieve desired information. Fairly frequently, they often encounter the problem of choosing the most appropriate terms to capture their information needs in their searches that match the terms in the underlying documents. As a feature provided by popular web search engines these days, users are aided through the tool of automatic query suggestion (QS). However, existing QS approaches are ineffective when users are looking for obscure documents. We facilitate this need by using the implicit structure of a corpus on which a search for desired information is performed. This is done by scoring potential suggested query lists using measures from information theory and recommending the suggested query list with the lowest score. The suggest queries retrieve the purest portions of the corpus in terms of user intent.