Cyberbullying detection and machine learning: a systematic literature review

The rise in research work focusing on detection of cyberbullying incidents on social media platforms particularly reflect how dire cyberbullying consequences are, regardless of age, gender or location. This paper examines scholarly publications (i.e., 2011–2022) on cyberbullying detection using machine learning through a systematic literature review approach. Specifically, articles were sought from six academic databases (Web of Science, ScienceDirect, IEEE Xplore, Association for Computing Machinery, Scopus, and Google Scholar), resulting in the identification of 4126 articles. A redundancy check followed by eligibility screening and quality assessment resulted in 68 articles included in this review. This review focused on three key aspects, namely, machine learning algorithms used to detect cyberbullying, features, and performance measures, and further supported with classification roles, language of study, data source and type of media. The findings are discussed, and research challenges and future directions are provided for researchers to explore.