Detection of Cyberbullying on User Tweets and Wikipedia Text using Machine Learning

On the internet, cyberbullying is a widespread issue that affects both adults and teenagers. Tragedies like suicide and sadness have resulted. On social media platforms, content regulation is becoming increasingly important. Based on hate speech tweets from Twitter and personal assault comments from Wikipedia forums, an effective model for identifying cyberbullying in text data is created. Data integration is performed to construct both tweets and Wikipedia datasets from multiple data sources separately. Feature extraction is performed using BoW and TF-IDF. SVM, Logistic Regressor, and Naive Bayes classifiers performance are evaluated on the integrated dataset. Out of the three classifiers, the SVM model is around 98.73% accurate on Twitter data and 95.46 % accurate on Wikipedia data.