其他摘要：The growth of the Arabic textual content on social media platforms has been caused by the continuous crises in the ArabWorld evoking the need to analyze the opinions of the public against the ongoing events. Arabic Sentiment Analysis (ASA) is, therefore, becoming the focus of many recent NLP studies. With several Arabic NLP resources being publicly available along with the emergence of deep learning techniques, researchers could handle the complex nature of Arabic language more efficiently. In the last decade, various ASA systems have been built. Yet, their achievements have not been investigated or compared against each other. This survey covers the ASA research carried out during the past five years. We compare and evaluate the performances and give insight into the ability of the created Arabic resources to support the future ASA research.
