• Gautam Prashantkumar Mehta
• Trupti Asolkar

Online social media and live interaction platforms are increasingly impacted by toxic and abusive language, making real-time content moderation essential. Early detection of toxic messages helps prevent cyberbullying, protect users, and maintain healthy online environments. Most existing approaches focus on a single language or platform, limiting their effectiveness in real-world, code-mixed scenarios. This work presents a web-based toxicity detection system using machine learning and deep learning techniques to analyze live chat and social media comments, including Hinglish and informal text. The system generates toxicity scores and moderation decisions using transformer-based models trained on publicly available datasets. An interactive dashboard visualizes real-time toxicity levels and actions, bridging the gap between research-oriented models and practical content moderation systems.

Real-Time Toxicity Detection, Cyberbullying Detection, NLP-based Text Analysis, Transformer Models, Hinglish Language Processing, Live Chat Moderation, Machine Learning, Content Moderation Dashboard.

"Real-Time Multilingual Toxicity Detection and Moderation in YouTube Live Chats", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.13, Issue 2, page no.c364-c369, February-2026, Available :http://www.jetir.org/papers/JETIR2602255.pdf

"Real-Time Multilingual Toxicity Detection and Moderation in YouTube Live Chats", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.13, Issue 2, page no. ppc364-c369, February-2026, Available at : http://www.jetir.org/papers/JETIR2602255.pdf