A property grammar-based method to enrich the Arabic treebank ATB
Résumé
We present a method based on the formalism of Property Grammars to enrich the Arabic treebank ATB with syntactic constraints (so-called proper-ties). The Property Grammar formalism is an effectively constraint-based ap-proach that directly specifies the constraints on information categories. This can facilitate the enrichment process. The latter is based on three phases: the problem formalization, the Property Grammar induction from the ATB and the treebank regeneration with a new syntactic property-based representation. The enrichment of the ATB can make it more useful for many NLP applications such as the am-biguity resolution. This allows also the acquisition of new linguistic resources and the ease of the probabilistic parsing process. This enrichment process is purely automatic and independent from any language and source corpus formal-ism. This motivates its reuse. We obtained good and encouraging experiment re-sults and various properties of different types.
Domaines
Traitement du texte et du document
Origine : Fichiers produits par l'(les) auteur(s)
Loading...