Acquisition de Connaissances à partir d’un texte Arabe non vocalisé

Abstract

The work we have done is the development of a toolbox for the acquisition of knowledge from an unvowelled Arabic text based on linguistic techniques for the automatic processing of natural language. The toolbox we developed called JEEM BOX. This work consists of three main parts: the first presents a state of art and provides the theoretical and methodological framework of our work and the second is devoted to the design and the implementation of JEEM BOX tools (the tools are: JStem for lemmatization, JTrans for transliteration, JClass for the classification of words based on their root, JConcord for the classification of words by their frequency, JDiac for vocalization and JExtract for the information extraction based on the different grammatical categories of the Arabic word). And the third part is for the experimental analysis of JEEM BOX tools.