A Human-Centered Systematic Literature Review of Cyberbullying Detection Algorithms

Cyberbullying is a growing problem across social media platforms, inflicting short and long-lasting effects on victims. To mitigate this problem, research has looked into building automated systems, powered by machine learning, to detect cyberbullying incidents, or the involved actors like victims perpetrators. In past, systematic reviews have examined approaches within body of work, but with focus computational aspects technical innovation, feature engineering, performance optimization, without centering around roles, beliefs, desires, expectations humans. paper, we present human-centered literature review past 10 years detection. We analyzed 56 papers based three-prong human-centeredness algorithm design framework - spanning theoretical, participatory, speculative design. found that fell incorporating multiple aspects, ranging from defining cyberbullying, establishing ground truth in data annotation, evaluating detection models, speculating usage users including potential harms negative consequences. Given sensitivities experience deep ramifications incidents bear actors, discuss takeaways how future can aid developing systems are more practical, useful, tuned diverse needs contexts stakeholders.