NLP Machine Learning for Online Harassment Detection on Social Media

Cyberbullying is a serious and growing concern that impacts both adults and teenagers, often leading to severe consequences such as depression and even suicide. As a result, there is an increasing need to regulate content shared on social media platforms. This study explores two distinct categories of cyberbullying: hate speech tweets from Twitter and personal attack comments from Wikipedia forums. The aim is to develop a detection model that can identify cyberbullying in textual data using natural language processing (NLP) and machine learning techniques. To determine the most effective approach, four different classifiers and three feature extraction methods are evaluated. The proposed model achieves over 90% accuracy on Twitter data and more than 80% accuracy on the Wikipedia dataset.