Breaking the Cyberbullying Cycle With Machine Learning

The proliferation of digital communication platforms has provided adolescents with unparalleled connectivity, yet it has concurrently amplified the menace of cyberbullying. This paper introduces "Serenity," a chat application prototype underpinned by advanced machine learning (ML) and natural language processing (NLP) techniques, designed to proactively identify and counteract cyberbullying incidents in real-time. Unlike traditional lexiconbased approaches, Serenity employs dynamic ML models, ensuring adaptability to the evolving linguistic patterns. The application not only detects potential cyberbullying but also incorporates mechanisms to deter such behaviors, thereby safeguarding the mental well-being of its young users. By integrating parental oversight without infringing on adolescents' privacy, Serenity strikes a balance between safety and autonomy. This research underscores the imperative of embedding sophisticated technological interventions in digital platforms catering to vulnerable populations and advocates for the broader adoption of such mechanisms in mainstream social media platforms. The design principles presented herein hold implications for platform developers, policymakers, and educators striving to consider in co-creating a safer digital ecosystem for adolescents.