Arabic Text Classification: A Literature Review

Abstract

Automatic text classification or categorization consists to assign predefined classes or categories to a given set of text documents aiming to organize the document collection based on conceptual views. Although there are many text classifiers in the literature, most of them are assessed using English or other non-Arabic languages text collections. The lack of availability of a large collection in the Arabic language is one of the most important challenges facing the few numbers of existing Arabic text classifiers (ATC). We present in this paper a literature review in the domain of Arabic text classification. We firstly overview the ATC based on machine learning algorithms. Then, we investigate ATC based on deep learning techniques as well as a set of other classifiers based on non-ML algorithms. The assessment of these ATC is also discussed. Finally, we focus on some open problems and we suggest some future directions.