Semantic Relations and Information Discovery

Cai, Di and van Rijsbergen, C.J. (2005) Semantic Relations and Information Discovery. In: Intelligent Data Mining: Techniques and Applications. Studies in Computational Intelligence, 5 . Springer, London, UK, pp. 79-102. ISBN 9783540262565

Preview

PDF - Accepted Version Download (253kB) | Preview

Abstract

The treatment of semantic relations between terms is essential in information retrieval (IR). Each term in a thesaurus might have classes of synonymous, narrower, broader, or related terms. The ability to express formally the semantic relations is a core issue in applying mathematical tools to IR. In the application of evidential theory, particularly, the problems become more apparent since evidential theory is based on set theory and individual terms have to be expressed as subsets of the frame of discernment. Four basic questions are to be faced: How to establish the frame of discernment using a thesaurus? How to express terms using subsets of the frame? How to apply evidential functions to represent documents or queries using the term subset-expressions? What are appropriate agreement measures for ranking documents against a given query? This study attempts to answer these questions.

Item control for Repository Staff only: