Arabic Morphological Analysis Based On Graphs And Correspondence tables Between Affixes And Root

摘要

In literature, there are several works on morphological analysis for the Arabic language. These works are based either on dictionaries of words or on dictionaries of stem words and also on the sum rules of the morphological analysis of this language, or on patterns of Arabic words. In this paper, we propose an improvement of the IBN-GINNY Morphological Analyzer I. We called it IBN-GINNY Morphological Analyzer II, in honor of the great Arabic linguist IBN-GINNY memory. This improvement combines the graphs with the Buckwalter morphological analysis approach. Furthermore, the approach uses a restricted dictionary and finds the analysis solutions in a global network via the Viterbi algorithm. However, in order to overcome certain drawbacks of the IBN-GINNY Morphological Analyzer I, our proposed approach integrates for the first time the correspondence tables between the prefixes infixes, infixes suffixes and infixes-roots within the IBN-GINNY system. Experiments were carried out on an Arabic corpus containing 10000 words. The obtained results are encouraging and show the importance of the proposed approach.