NLP DRIVEN CYBERBULLYING DETECTION SYSTEM FOR SOCIAL MEDIA USING TRANSFORMED BASED SENTIMENTAL ANALYSIS

Abstract

The rapid growth of social media platforms has
significantly transformed digital communication by
enabling users to share information, opinions, and
multimedia content globally. However, this
widespread connectivity has also increased the
occurrence of cyberbullying, online harassment,
abusive comments, and toxic interactions that
negatively affect individuals, particularly teenagers
and young users. Cyberbullying can lead to severe
psychological, emotional, and social consequences
including anxiety, depression, stress, and suicidal
tendencies. Manual monitoring of harmful content
across millions of social media posts is highly
inefficient and impractical due to the enormous
volume of user-generated data. Therefore, an
automated and intelligent cyberbullying detection
system is essential for ensuring safer online
communication. This project presents an NLP
Driven Cyberbullying Detection System for Social
Media Using Transformer Based Sentimental
Analysis that combines Natural Language
Processing techniques with machine learning
algorithms to identify abusive and bullying content
effectively. The system performs text preprocessing
operations such as tokenization, stop-word
removal, normalization, and stemming to clean
noisy social media data. Feature extraction
techniques including Bag-of-Words and TF-IDF are
applied to convert textual information into
numerical vectors suitable for classification.
Transformer-based sentiment analysis is
incorporated to capture contextual meaning and
semantic relationships within messages, improving
the detection of implicit and sarcastic bullying
content. Multiple machine learning classifiers such
as Support Vector Machine, Logistic Regression,
and Naive Bayes are evaluated to achieve high
classification accuracy. The proposed system
provides automated monitoring, scalable
deployment, improved text representation, and
enhanced detection performance for real-time
cyberbullying identification. The system ultimately
contributes to creating a safer digital environment
and supports social media moderation through
intelligent harmful content filtering.