Handling Structural Ambiguity in a Knowledge-Based Information Retrieval System

This paper presents a strategy to handle syntactic ambiguity in a theoretically motivated fashion following general linguistic principles. This strategy, which is called underspecification, was implemented in a Natural Language Engine (NLE) for automatic information extraction, called ELSA (an acronym for English Language Semantic Analyser), which was developed at the Department of Language & Speech of the University of Nijmegen. The crucial idea of the strategy is that, in case of ambiguity, the NLE should know what option to choose and when to choose it. Until that moment the analysis remains underspecified, i.e. only one derivation is produced. At present time, the NLE in question is adapted to serve as the linguistic module of a knowledge-based Information Retrieval System, called Condorcet, being developed at theUniversity of Twente, for documents on the fields ofmechanical properties of engineering ceramics as a subfield of engineering, and epilepsy as a subfield of medicine. In this paper we will show how a theory-driven NLE will make a substantial contribution to (semi)automatic information retrieval, making use of the AGFL system. The authors are greatly indebted to Nicolaas J.I. Mars and Paul E. van der Vet, the initiators of the Condorcet project, for their substantial contribution to this article. Chapter