CYBERBULLYING DETECTION ON SOCIAL NETWORKS COMPARING MACHINE LEARNING AND TRANSFER LEARNING

Abstract

Communication has been transformed by information and communication technologies, yet cyberbullying still presents significant difficulties that need for automated solutions for efficient detection on social media platforms. The PROJECT integrates both
conventional Machine Learning and Transfer Learning methodologies, emphasizing feature extraction and selection strategies to improve the model's comprehension of cyberbullying situations. For reliable cyberbullying detection, the project makes use of a variety of models, including as LinearSVC, Logistic Regression, DistilBert, DistilRoBerta, and Electra. These models also use Machine Learning, Transfer Learning, and Deep Learning techniques.