Cyberbullying detection on Bangla social media comments

Abstract

This study investigates the potential of machine learning algorithms in detecting signs of depression within Bengali text on social media platforms. As mental health concerns continue to rise globally, understanding linguistic patterns indicative of depression in the unique context of the Bengali-speaking population becomes imperative. Leveraging natural language processing, the study employs a variety of machine learning algorithms, including Support Vector Machine (SVM), Random Forest, Long Short-Term Memory (LSTM), and Bidirectional Long Short-Term Memory (Bi-LSTM). Ethical considerations take precedence throughout the study, focusing on user privacy, informed consent, and cultural sensitivity. The research aims not only to develop effective depression detection models but also to ensure responsible data governance and user empowerment. By fostering a supportive environment for mental health discussions, the study aligns its objectives with ethical principles. Results showcase promising accuracy rates, with SVM leading at 86%, followed by LSTM at 83%, Bi-LSTM at 81%, and Random Forest at 79%. Beyond accuracy, the study evaluates precision, recall, and F1-score metrics to provide a comprehensive understanding of each algorithm's performance. The implications of the research extend beyond numerical metrics. The study advocates for the development of culturally sensitive and user-centric interventions, emphasizing the importance of ethical considerations in deploying artificial intelligence for mental health support. In the Bengali-speaking community, where cultural nuances play a significant role in linguistic expressions, the study's outcomes contribute valuable insights for tailoring depression detection tools to local contexts.