Search

The increasing use of social media has led to the emergence of a new challenge in the form of abusive content. There are many forms of abusive content such as hate speech, cyberbullying, offensive language, and abusive language. This article will present a review of abusive content automatic detection approaches. Specifically, we are focusing on the recent contributions that were using natural language processing (NLP) technologies to detect the abusive content in social media. Accordingly, we adopt PRISMA flow chart for selecting the related papers and filtering process with some of inclusion and exclusion criteria. Therefore, we select 25 papers for meta-analysis and another 87 papers were cited in this article during the span of 2017–2021. In addition, we searched for the available datasets that are related to abusive content categories in three repositories and we highlighted some points related to the obtained results. Moreover, after a comprehensive review this article propose a new taxonomy of abusive content automatic detection by covering five different aspects and tasks. The proposed taxonomy gives insights and a holistic view of the automatic detection process. Finally, this article discusses and highlights the challenges and opportunities for the abusive content automatic detection problem.

Different social media platforms have become increasingly popular in the Arab world in recent years. The increasing use of social media, however, has also led to the emergence of a new challenge in the form of abusive content, including hate speech, offensive language, and abusive language. Existing research work focuses on automatic abusive content detection as a binary classification problem. In addition, the existing research work on the automatic detection task surrounding abusive Arabic content fails to tackle the dialect-specific phenomenon. Consequently, this has led to two important issues in the automatic abusive Arabic content detection task. In this study, we used a multi-aspect annotation schema to tackle the automatic abusive content detection problem in Arabic countries, based on the multi-class classification task and the dialectal Arabic (DA)-specific phenomenon. More precisely, the multi-aspect annotation schema includes five attributes: directness, hostility, target, group, and annotator. We specifically developed a framework to automatically detecting abusive content on Twitter using natural language processing (NLP) techniques. The developed framework used different models of machine learning (ML), deep learning (DL), and pretrained Arabic language models (LMs) using the multi-aspect annotation dataset. In addition, to investigate the impact of the other approaches, such as multi-task learning (MTL), we developed four MTL models built on top of a pretrained DA language model (called MARBERT) and trained on the multi-aspect annotation dataset. Our MTL models and pretrained Arabic LMs enhanced the performance compared to the existing DL model mentioned in the literature.