Enriching a lexicographic tool with domain definitions: Problems and solutions

Resumen

Enriching linguistic resources with domain information has been considered one important target in natural language applications. However, automatic definition extraction of this domain information from specialized resources has revealed certain methodological problems in definition construction. This paper presents some problems encountered in automatic definition extraction that are mainly related to inconsistencies in definitions, different granularity of definitions and embedded definitions. To face these problems some Meaning-Text Theory tools have been used: (a) semantic labels as a solution for inferring knowledge, (b) lexical functions as a way of providing coherence to definitions and (c) the actancial structure as a tool for developing consistent and complete definitions. Our goal is to describe the problems and to show the solutions proposed.