La reconnaissance automatique de la fonction des pronoms démonstratifs en langue arabe (Automatic recognition of demonstrative pronouns function in Arabic) [in French]

________________________________________________________________________________________________________ Automatic recognition of demonstrative pronouns function in Arabic Anaphora resolution is one of the most difficult tasks in NLP. Classifying pronouns before attempting a task of anaphora resolution is important because to handle the cataphoric pronoun, the system should determine the antecedent into the segment following the pronoun. Although, for the anaphoric pronoun, the system should look for the antecedent into the segment before the pronoun. In addition, the number of demonstrative pronouns is very important in Arabic. In this paper, we describe a machine learning method for classifying demonstrative pronouns in Arabic. We have evaluated our approach on a corpus of 365585 words which contain 14318 demonstrative pronouns and we have obtained encouraging results: 99.3% as F-Measure. MOTS-CLES : Pronoms démonstratifs, résolution des anaphores, traitement de la langue arabe.