Social Media Toxic Comment Classifier

Abstract

The rapid growth of social media platforms has enhanced communication but also increased 
the spread of toxic and abusive content. This project develops a social media toxic comment 
classifier using natural language processing (NLP) and machine learning to automatically 
detect and categorize comments as toxic, sever toxic, obscene,  threat,  insult,  or  identity 
hate. Using the jigsaw Toxic Comment Classification Challenge dataset, the system performs 
data preprocessing, tokenization, feature extraction, and classification through algorithms like 
Logistic Regression, Naïve Bayes, and deep learning models. The classifier achieves high 
accuracy and can be integrated into online platforms to promote healthier digital interactions, 
helping reduce online toxicity and foster safer, more inclusive communities.