A Comparative Analysis on AI-Driven Speech Protection Approaches

Automatic hate speech detection is a crucial tool for mitigating the spread of hate speech on social networking sites. This study compares and evaluates various shallow and deep learning methods using three popular hate speech datasets. We aim to assess the current state of the art, identify its strengths and weaknesses, and determine its future direction. Key performance indicators such as detection accuracy, computation speed, pre-trained model utilization, and domain generalization are analyzed.