International Journal of Scientific Research and Engineering Development

The rapid expansion of online communication platforms has led to a noticeable rise in offensive language, hate speech, and cyberbullying. Such harmful content affects users emotionally and contributes to unsafe digital spaces. Conventional content moderation methods, including keyword-based filtering and manual supervision, often fail to identify offensive messages that rely on context, implicit meaning, or indirect expressions. To overcome these limitations, this paper presents Offensive-Bot, a real-time offensive message detection system developed using Natural Language Processing (NLP) and a fine-tuned Bidirectional Encoder Representations from Transformers (BERT) model. The system examines user messages before they are sent and categorizes them as either safe or offensive. When offensive content is detected, the message is immediately blocked and the user is alerted, encouraging responsible online behavior. In addition, restricted messages are stored for administrative monitoring and future system improvement. Experimental results indicate that the proposed system achieves high detection accuracy with minimal processing delay. Overall, OffensiveBot contributes to creating safer and more respectful online communication environments.

Ms.P.Revathy, K.Jayashree, R.Preethaa,"Offensive-Bot: BERT and NLP based Real-Time Offensive Message Detection and Restriction System" International Journal of Scientific Research and Engineering Development, V9(2): Page(242-246) Mar-Apr 2026. ISSN: 2581-7175. www.ijsred.com. Published by Scientific and Academic Research Publishing.