Respectful Social Engagement: Facebook Hate Speech Classification for Bengali Language

In today's digital landscape, the proliferation of hate speech and cyberbullying on social media platforms like Facebook has become a societal concern due to its impact on individuals and communities. The increase in suicides and political, personal, and religious violence in Bangladesh highlights the urgent need to combat hate speech, especially in Bengali, which is common in Facebook comments. With the specific goal of creating a robust algorithm to identify and categorize such speech, this study focuses on identifying hate speech in Bangla in Facebook comments. In order to offer a comprehensive knowledge of hate speech dynamics within the Bengali-speaking community, the research makes use of a primary diverse dataset that consists of 4,234 social media comments that contain a wide range of regional Bengali vocabulary. The study aims to distinguish between neutral and inflammatory language patterns using a variety of machine learning methods, such as Support Vector Machines (SVM), Decision Trees (DT), Random Forests (RF), Logistic Regression (LR), and K-Nearest Neighbors (KNN). Notably, SVM emerged as the most accurate classifier, which achieves a remarkable accuracy rate of 96%. Focusing on the unique difficulties of identifying hate speech in Bengali, especially given its regional linguistic variances, is what makes this work novel. Additionally, extensions for further research are also included.