Paper Title
Machine Learning Applied on Arabic Text Diacritization: Survey

Abstract
Arabic language is the fourth most used language on the Internet. While the number of different Arabic corpora continues to grow, there is an increasing need for robust tools to process that data, for many essential purposes of Natural Language Processing (NLP). In Arabic language, texts are often written without diacritic marks. However, those marks are important to clarify the sense and meaning of words and the lack of them lead to ambiguity even for the natives. Unfortunately, many Arabic NLP tools such as machine translation, sentiment analysis, and text-to-speech are vulnerable due to the lack of diacritical marking. Automatic diacritization is the process of restoring diacritic marks. In this study, we present challenges that Arabic diacritization tools have to overcome. Then we overview the existing diacritization tools and we focus on Marching Learning (ML) methods. Afterwards we survey recent works that address the diacritization problem using ML algorithms. Finally, we synthesize the results of the studies, identify research gaps, and offer recommendations for researchers working in this area. Keywords - Arabic text Diacritization, Machine Learning, Neural Networks, Arabic Natural language Processing, Deep Learning