Cyberbullying in Students

Abstract

This literature review summarizes twenty research articles on many of the facets of online risks, cyberbullying and harassment, trolling, and digital safety awareness. Cyberbullying, online harassment, trolling and other online risks have emerged as a global societal challenge affecting mental health, education and socialization. Each of the articles is based on a diverse set of data sources (expert assessments, surveys on social media, questionnaires from adolescents, tweets, and, experiments utilizing games) and employs a variety of methods, including conceptual frameworks and psychosocial analysis, statistical modelling approaches, and computational methods using modern machine learning techniques. The computational approaches use various methods, including MultiCriteria Multi Decision Maker (MCMDM) approaches (like Fuzzy Analytic Hierarchical Process - AHP and Game Theory), transformers (ELECTRAPOS designed with part-of-speech fusion features), traditional classifiers (Support Vector Machine (SVM), Logistic Regression (LR), and Naive Bayes (NB) especially suited with TF-IDF, N, gramming, and sentiment and emotion features). Other summary classification approaches used FastText, Word2Vec embeddings, and intention detection measures, delivering strong predictive ability to find the highest detection accuracy of 96.9% in identifying cyber harassment. Other authors explained that deepfake risks involved emotional manipulation and identity impersonation were utilized, trolling was based on anonymity and conformity by social contagion within a group, and sexting and romantic myths from adolescents were all correlated to cyber abuse. Directiveintent levels in policy were introduced through interventions in how mobile phone bans adopted in schools demonstrated both advantages, (engagement, improvements in social interaction) and disadvantages (a decline in independence or accessibility of digital tool use); while game-based or inserted-gaming approaches were demonstrated to use awareness for building up responsiveness to cybersecurity in-game interactions and entity session-based detection frameworks reflected on the detection of prior abuse log entries, as based on how context and repetition in abuse form within or outside the session. This collection of research articles reflected off the multidisciplinary nature of research that explicates the risks and harm of cyberspace, and suggests issues of integrative strategies across levels of psychology, socialization, and computational understandings to develop prevention, detection, and resilience in cyberspace.