SecureComment: Safeguarding Online Discussions with Intelligent Toxic Comment Filtering
Reddy Kowshik Rayani, Samhitha Tekula, Subhash Kovid Vattigunta, Naveen Kumar Kovi, Kalakunnath Namitha · 2024
Toxic comments, such as hate speech and abuse, are a widespread issue online, disrupting healthy conversation and user safety. Identifying and filtering these comments is crucial for ensuring respectful online communications. The primary challenge arises from the need to obtain accurate detection while addressing emerging tactics and balancing between false positives and negatives. This concern affects user interactions and platform credibility and raises legal and ethical risks. Therefore, this study examines the effectiveness of numerous machine-learning models and implements an ensemble method for analyzing and classifying toxic comments. We analyze the effectiveness of various algorithms using evaluation metrics like precision, recall, accuracy, and F1-score. Ensemble methods are utilized to combine the capabilities of several models, enhancing the classification of toxic comments. The study aspires to deliver insights into the comparative performance of various machine learning models in classifying toxic comments accurately and the advantages of ensemble methods in enhancing classification accuracy.