The use of relation matching in information retrieval

Abstract

Most information retrieval systems use keywordmatching to identify documents that may be relevant to the user's query. The user's query may contain multiple concepts linked together with relations. However, keyword matching methods ignore relations that are expressed in the query. This paper discusses the use of relation matching in information retrievalmatching terms and relations between terms expressed in the query with terms and relations found in the documents. Previous studies on relation matching are surveyed, and related issues and problems are discussed. The results of research carried out as part of the author's dissertation are reported and current research efforts are outlined. This document may be circulated freely with the following statement included in its entirety:

Phase-based information retrieval

“…For an excellent literature review on these subjects we refer the reader to the thesis work of Khoo, 1997. However, we choose to review three studies which we consider representative and close-related to the approach we are going to suggest in this article.…”

Multi-word terms selection for information retrieval

“…It is difficult to place MWT at a semantic level. In a practical way, it is the syntactic structure that serves as a gateway to the semantic level (Khoo, 1997). Indeed, Carballo and Strzalkowski (2000) indicate that linguistic processing, for a performance of documents with MWT, can cover, unlike a representation with simple terms, some aspects of the semantic content of documents.…”