PredXtract, a generic platform to extract in texts predicate argument structures
Résumé
Verbal and nominal predicate structures present interesting properties for information extraction. We show how to study these predicate structures in a uniform way, using the fact that the nominalization of a verb has the same arguments as the verb. We then describe the extraction platform (PredXtract) which we have developed in order to extract predicate argument structures and which highlights relations between biological entities in biological texts. We present and discuss our results.
Domaines
Traitement du texte et du document
Origine : Fichiers produits par l'(les) auteur(s)
Loading...