talnarchives

Une archive numérique francophone des articles de recherche en Traitement Automatique de la Langue.

An Example-Based Semantic Parser for Natural Language

Michel Généreux

Abstract : This paper presents a method for guiding semantic parsers based on a statistical model. The parser is example driven, that is, it learns how to interpret a new utterance by looking at some examples. It is mainly predicated on the idea that similarities exist between contexts in which individual parsing actions take place. Those similarities are then used to compute the degree of certainty of a particular parse. The treatment of word order and the disambiguation of meanings can therefore be learned.

Mots clés : Analyseur sémantique, Corpus, Langue naturelle

Keywords : Semantic parser, Corpus, Natural language