Text Classification of Arabic Text: Deep Learning in ANLP

Abstract

Recently Natural Language Processing (NLP) has excessive attention due to increased data available online and needs processing. Nevertheless, the huge development in the NLP but still Arabic Natural Language Processing (ANLP) faces many challenges and grief from researchers’ leakage compared with English NLP. Mainly this study has three divisions. It aims to find out the challenges facing ANLP, especially text classification.Furthermore, it aims to examine the effect of using deep learning in ANLP text classification from both sides’ advantages and disadvantages. Moreover, it aims to find the most efficient deep learning algorithm that sufficiently classifies and categorizes an Arabic text. To fulfill the objectives of this study, three different methods have been used. The first method, Survey / Systematic Review, to find out the challenges. The second method, Literature review; to examine the effect of using deep learning algorithms to classify Arabic text. The third method, experiment to find an efficient algorithm that reverts better accuracy. In this study, we have determined the data collection methods. Where searching for good quality research papers is the key to accomplish the first two divisions. But the last division needs the full implementation to deploy Arabic text classification models. The Data Analysis explained the sub-steps extensively in each division, how it will be implemented, and each division’s expected outcome. Also, this study briefly gives the resources needed and the required timeline. Deploying the models will be the longest process; hence it needs preprocessing and building models covering all types of neural networks to determine the most efficient one.KeywordsArabic Natural Language Processing (ANLP)Deep learningArabic text classification