International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 12 Issue: 04 | Apr 2025
p-ISSN: 2395-0072
www.irjet.net
ChatEclipse360: Intelligent Toxic Concealment System Chakka Naga Venkata Satya Sai Siri1, Gonnuri Lalitha Sai Jayamani2, Paidikondala Sarasvathi Sai Himaja3, Ms. Ch Naga Padma Latha4 1 CST, Sri Vasavi Engineering College(A), Pedatadepalli, Tadepalligudem – 534101. 2 CST, Sri Vasavi Engineering College(A), Pedatadepalli, Tadepalligudem – 534101. 3 CST, Sri Vasavi Engineering College(A), Pedatadepalli, Tadepalligudem –534101.
4 Assistant Professor, Department CSE, Sri Vasavi Engineering College(A), Pedatadepalli, Tadepalligudem-534101
---------------------------------------------------------------------***--------------------------------------------------------------------while still notifying the sender. This feature helps educate Abstract - The rise of online communication has led to an
users about respectful communication and prevents disruptions in conversation flow.
increased prevalence of toxic messages that negatively impact user experiences. Our project focuses on developing an intelligent system to identify and mitigate toxic messages within chat applications. The proposed solution incorporates Natural Language Processing (NLP) and machine learning techniques for detecting and masking toxic content in realtime. When a sender attempts to send a message containing toxic words or phrases, the system immediately intervenes by masking the message. The receiver will see a placeholder indicating the message was identified as toxic, without revealing the specific toxic content. This preserves transparency without propagating harmful language. The sender receives a notification indicating that the message contains inappropriate content, encouraging more respectful communication behavior. This project aims to foster healthier online interactions by preventing the circulation of offensive or harmful language. Our solution is highly scalable, ensuring seamless integration into existing chat applications, while maintaining user privacy and optimizing performance
By implementing this solution, we aim to create a safer and more positive digital space where users can engage in meaningful and respectful conversations without the fear of encountering harmful content.
2. LITERATURE SURVEY 2.1) Navoneel Chakrabarty.“A Machine Learning Approach to Comment Toxicity Classification” (2020). AI Research Journal. This study explores the use of machine learning techniques to classify toxic content based on categories such as threats and identity-based abuse obscenity, threats, insults, and identity-based hatred, to filter harmful content.[1] 2.2 H. Masoorian, M. Ahmadi, N. Mohammadzadeh, and S. M. Ayyoubzadeh. “Research of Techniques used in Toxicity Detection” (2021). International Journal of Artificial Intelligence. They proposed the use of neural networks and machine learning techniques for automatic detection of toxic comments, aiming to protect users from harmful online behavior.[2]
Key Words: BERT, Concealment System, Content moderation, Real -time Chat Security, ML, NLP, Real-time moderation, Respectful communication, Toxic message detection, User safety, Message filtering, NLP-based moderation, Offensive language detection
2.3 P. G. Davange, P. Chaudhari, S. T. Patil, and A. Bhojawala.” Toxic Chat Detection using Deep Learning. (2023)”. They proposed the use of machine learning and deep learning, particularly LSTM with BERT word embeddings, to categorize and filter toxic comments, achieving 94% accuracy.[3]
1. INTRODUCTION In today's digital era, online communication platforms have become an integral part of daily interactions. However, the widespread use of these platforms has led to an increase in toxic and harmful messages, negatively impacting user experiences and fostering an unsafe digital environment.
2.4 D. Nithya, Nanthine K.S., Thenmozhi S., & Varshini Priya R. “Advanced social media Toxic Comments Detection System Using AI” (2024). They proposed the development of an automated system for detecting and flagging toxic comments in real-time on social media platforms using NLP and ML techniques.[4]
To address this issue, our project, ChatEclipse360, introduces an Intelligent Toxic Message Concealment System designed to detect and moderate harmful content in realtime. Leveraging the power of Natural Language Processing (NLP) and Machine Learning, this system identifies toxic messages before they are delivered, ensuring that inappropriate content does not propagate further. Unlike traditional moderation tools that either completely block messages or rely on manual intervention, ChatEclipse360 takes a more refined approach by masking harmful messages
© 2025, IRJET
|
Impact Factor value: 8.315
2.5 Z Yang, D Tullo, R Rabbany.” ToxiSight: Insights Towards Detected Chat Toxicity” (2024).. They propose an explainability dashboard for in-game chat toxicity detection, integrating XAI techniques like token importance analysis, model visualization, and dataset attribution. The
|
ISO 9001:2008 Certified Journal
|
Page 1390