Deep learning for Arabic subjective sentiment analysis: Challenges and research opportunities

The fields of machine learning and Web technologies have witnessed significant development in the last years. This caused a ceaseless and rapid growth in sharing of the views and experience regarding services or products over the Internet in different domains. Therefore, a torrential flow of online data is available for analytical studies. Sentiment Analysis (SA) is a subtask of Natural Language Processing (NLP) that aims to analyze huge data for detecting people opinions and emotions. This field has gained growing interest by public and private sectors that led to the occurrence of many challenges, especially that related to the Arabic language. The purpose of this study is to conduct a systematic review from year 2000 until June, 2020 to analyze the status of deep Learning for Arabic NLP (ANLP) task in Arabic Subjective Sentiment Analysis (ASSA) to highlight the challenges and propose research opportunities in this field. Extensive number of research studies were reviewed to investigate deep learning techniques applied in subjective sentiment analysis for the Arabic language. We observed that CNN and RNN (LSTM) models were the most common methods used for ASSA. The results of this review show that there is a need for more efforts to implement modernized deep learning methods for Arabic sentiment analysis systems. (C) 2020 Elsevier B.V. All rights reserved.

Automated summary

The fields of machine learning and web technologies have seen significant development in recent years, leading to a vast amount of online data available for analytical studies. Sentiment Analysis, as a subtask of Natural Language Processing, aims to analyze opinions and emotions. A systematic review on the application of deep learning techniques for subjective sentiment analysis in Arabic language reveals the need for more efforts to implement modernized methods.