HMM Based Approach for Arabic Word Segmentation

Word decomposition or segmentation is one of the first stages of Natural Language Processing, particularly written language. It aims to find the decomposition of each compound word into its sub-elements. However, given the inflectional nature of the Arabic language, this process becomes more difficult for Arabic words, especially for ambiguous cases, in which the same word can have several possible decompositions. In this article, we try to propose a new approach based on the Hidden Markov Model to decide which segmentation to choose in case of ambiguity. The decision will be based on an auxiliary element predicted by the HMM for each word according to its context.