English vs Arabic Sentiment Analysis: A Survey Presenting 100 Work Studies, Resources and Tools

Abstract

To determine whether a document or a sentence expresses a positive or negative sentiment, three main approach types are commonly used: lexicon based approaches, machine learning (ML) based approaches and hybrid approaches. English has the greatest number of sentiment analysis studies, unlike other languages including Arabic and its dialects. More specifically, ML based sentiment analysis requires annotated data. In the case of lexicon based approaches, they typically require the availability of lexicons annotated by valence and/or intensity. One of the majors problems related to the treatment of Arabic and its dialect is the lack of the above resources. This survey is aimed to highlight the most recent resources that have been constructed and most recent advances in the context of sentiment analysis (related to English and Arabic language). It refers to about one hundred recent papers, most of which published between 2015 and 2018. These works are classified by category (as survey work vs. solution work). In the case of solution works, we focus on the construction of sentiment lexicon and corpus. We also propose new trends in Arabic sentiment analysis, mainly employing deep learning techniques.