Deep Neural Networks for Detecting Hate Speech

Abstract

With increased social media activity, hate speech on the Internet has become increasingly prevalent. It is a hazardous and damaging form of internet content that targets a group or individual based on their religion, race, or sexual orientation. As a result, it has garnered increasing attention from researchers. This article will examine current research trends, data sources, and methodologies and recommend future study directions. The subject of hate speech was determined at the start of 2020, and it includes attacks on minorities, religion, women, the general election agenda, and politics. Individuals have experimented with various methodologies and models and discovered several characteristics that fulfill the exclusion requirements. However, these methodologies and features do not imply that they will perform well in detecting hatred. The data collection, selected features, number of categories, and mutually exclusive categories all significantly impact the classification performance of hate speech.