This document discusses using machine learning approaches to detect cyberbullying on social networks. It begins with an abstract that introduces the problem of cyberbullying and its negative impacts. It then discusses related work on using natural language processing and machine learning like SVM, neural networks, and co-training models to identify cyberbullying. The document aims to compare different machine learning algorithms (Naive Bayes, Decision Tree, Random Forest, SVM, DNN) to determine the most effective technique for detecting cyberbullying. It discusses preprocessing data and extracting features before applying the algorithms. The results will show which algorithm has the highest accuracy on a Twitter comment dataset.