Conception d'un jeu de ressources libres pour le TAL arabe sous Unitex - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2013

Conception d'un jeu de ressources libres pour le TAL arabe sous Unitex

Résumé

This paper aims to describe the process of building a free Arabic package for the Unitex framework: we proposed a test corpus, we chose a tag set suited to this task and we build dictionaries respecting the LADL DELA format. We describe each of the above particularly the building of dictionaries, for which we designed algorithms for automatic generation of verb and noun inflection graphs. We use the word-based inflection foundations and we define for each lexeme a set of themes. For the verbs, we use five themes given by the user and the graphs generate up to 264 inflected verbal forms; for the nouns and adjectives we use one or at most two themes and the produced graphs generate 63 inflected forms.
Fichier non déposé

Dates et versions

hal-01024409 , version 1 (16-07-2014)

Identifiants

  • HAL Id : hal-01024409 , version 1

Citer

Noureddine Doumi, Ahmed Lehireche, Denis Maurel, Moussa Ali Cherif. Conception d'un jeu de ressources libres pour le TAL arabe sous Unitex. 6ème colloque international en Traductologie et TAL, 2013, Oran, Algérie. ⟨hal-01024409⟩
145 Consultations
0 Téléchargements

Partager

Gmail Facebook X LinkedIn More