COMPARATIVE ANALYSIS OF ONLINE ABUSE DETECTION TECHNIQUES: TRADITIONAL MACHINE LEARNING, DEEP LEARNING, AND HYBRID MODELS

Abstract

Abstract: Online abuse is a prevalent issue on social media platforms, necessitating the development of effective detection techniques. This study compares traditional machine learning models (Naive et al.), advanced deep learning models (CNN, RNN, LSTM, BERT, GPT), and hybrid models that combine these approaches. Utilizing diverse datasets from platforms such as Twitter, Reddit, Kaggle, Wikipedia, YouTube, and Facebook, the research evaluates model performance using metrics like accuracy, precision, Recall, F1-Score, specificity, AUC-ROC, AUC-PR, and MCC. The findings highlight that deep learning models outperform traditional models, particularly transformer-based models like BERT. Hybrid models also show improved robustness and adaptability, making them effective against evolving online abuse. This study underscores the importance of adopting advanced, adaptable, and thoroughly evaluated models to enhance online abuse detection and promote safer online environments.