Detecting Digital Harassment : The Bullynet Approach To Cyberbully Identification (Nlp)

Abstract

The rapid growth of digital communication platforms has transformed how people interact, but it has also led to an increase in cyberbullying and online harassment, creating significant challenges for users, especially adolescents and young adults. While existing moderation systems attempt to address this issue, their effectiveness remains limited due to reliance on manual reporting and keyword-based filtering, which fail to capture contextual meaning and evolving slang. This paper presents BullyNet, a system that detects cyberbullying in textual data using Natural Language Processing (NLP) techniques. It combines transformer-based models such as BERT with sentiment analysis and hybrid classification to improve detection accuracy. The system processes user -generated content, identifies harmful intent, and classifies abusive language while reducing false positives. Evaluated for accuracy and efficiency using datasets from platforms like Twitter and YouTube, the system shows promise as an effective tool to enhance online safety and promote healthier digital communication environments.