I. Ben Cheikha, A. Kacema and A. Belaïdb
aUTIC-ESSTT, 5 Avenue Taha Hussein, BP56 Bab Menara, 1008 Tunis, Tunisie;
bLORIA, Campus scientifique
 

Abstract
This paper introduces a novel approach for the recognition of a wide vocabulary of Arabic words. Note that there is an essential difference between global and analytic approaches in pattern recognition. While the global approach is limited to reduced vocabulary, the analytic approach succeeds to recognize a wide vocabulary but meets the problems of word segmentation especially for Arabic. We have investigated the use of Arabic linguistic knowledge to improve the recognition of wide Arabic word lexicon. A neural-linguistic approach was proposed to mainly deal with canonical vocabulary of decomposable words derived from tri-consonant healthy roots. The basic idea is to factorize words by their roots and schemes. In this direction, we conceived two neural networks TNN_R and TNN_S to respectively recognize roots and schemes from structural primitives of words. The proposal approach achieved promising results. Enlarging the vocabulary from 1000 to 1700 by 100 words, again confirmed the results without altering the networks stability.

Keywords: Arabic Writing Recognition, Wide Canonical Vocabulary, Neural Networks, Structural Primitives, Linguistic Knowledge Integration