How effective is query expansion for finding novel information?

Authors:
Min Zhang;Chuan Lin;Shaoping Ma
Affiliations:
State Key Lab of Intelligent Tech. and Sys, Tsinghua University, Beijing, China;State Key Lab of Intelligent Tech. and Sys, Tsinghua University, Beijing, China;State Key Lab of Intelligent Tech. and Sys, Tsinghua University, Beijing, China
Venue:
IJCNLP'04 Proceedings of the First international joint conference on Natural Language Processing
Year:
2004

Citing 9
Cited 0

Experiments in automatic statistical thesaurus construction

SIGIR '92 Proceedings of the 15th annual international ACM SIGIR conference on Research and development in information retrieval
Experiment on linguistically-based term associations

Information Processing and Management: an International Journal
Query expansion using lexical-semantic relations

SIGIR '94 Proceedings of the 17th annual international ACM SIGIR conference on Research and development in information retrieval
Automatic thesaurus generation for an electronic community system

Journal of the American Society for Information Science
Query expansion using local and global document analysis

SIGIR '96 Proceedings of the 19th annual international ACM SIGIR conference on Research and development in information retrieval
Local Feedback in Full-Text Retrieval Systems

Journal of the ACM (JACM)
Explorations in Automatic Thesaurus Discovery

Explorations in Automatic Thesaurus Discovery
Concept discovery from text

COLING '02 Proceedings of the 19th international conference on Computational linguistics - Volume 1
Identifying synonyms among distributionally similar words

IJCAI'03 Proceedings of the 18th international joint conference on Artificial intelligence

Quantified Score

Hi-index	0.00

Visualization

Abstract

The task of finding novel information in information retrieval (IR) has been proposed recently and paid more attention to. Compared with techniques in traditional document-level retrieval, query expansion (QE) is dominant in the new task. This paper gives an empirical study on the effectiveness of different QE techniques on finding novel information. The conclusion is drawn according to experiments on two standard test collections of TREC2002 and TREC2003 novelty tracks. Local co-occurrence-based QE approach performs best and makes more than 15% consistent improvement, which enhances both precision and recall in some cases. Proximity-based and dependency-based QE are also effective that both make about 10% progress. Pseudo relevance feedback works better than semantics-based QE and the latter one is not helpful on finding novel information.