Extracción de relaciones sintagmáticas de corpus anotados

  1. Navarro Colorado, Borja
  2. Moreno Monteagudo, Lorenza
  3. Martínez Barco, Patricio
Journal:
Procesamiento del lenguaje natural

ISSN: 1135-5948

Year of publication: 2006

Issue: 37

Pages: 67-74

Type: Article

More publications in: Procesamiento del lenguaje natural

Abstract

In this paper, we present a new resource, designed for being used in WSD, based on syntagmatic relations between senses for Spanish. These relations have been extracted from a corpus: the Cast3LB corpus which has been manually annotated with syntactic and semantic information (WordNet senses). From it, approximately 3000 patterns have been extracted. These patterns show the syntagmatic relations between verb senses and its arguments within a sentence. However, these patterns can be too specific to be used in multilingual contexts or in open domain texts. Consequently, it is necessary to obtain more abstract patterns. In order to do so, we have also developed general patterns using semantic classes based on the SUMO ontology.