Automatic and Advance Techniques for Hate Speech Detection on Social Media: A Review

The aim of the study is to review automatic and advanced techniques for investigating hate and offensive speech from social media (SM) platforms. Finding hateful speech from social media is a text classification problem. In the proposed paper explains the methodology of automatic text classification through the medium of traditional machine learning and advanced deep learning algorithms. On social media, people share their opinion and different content, but some users post hateful and offensive content. Detecting and classifying hate speech from social sites is not a small challenge. There are simply five steps that are collecting the data, data cleaning and pre-processing, applying feature extraction techniques, training and testing data in the classification algorithm, and comparative analysis of the algorithm's performance. This review, analyzes the performance of the confusing metrics concepts using four metrics precision (Pr), recall (Re), F1-score, and accuracy (A). Role of this study is to update the researchers and readers on the state-of-the-art model and technology for hateful speech classification. In the last of, this review paper explains some challenges and research gaps for identifying the hate speech in existing models.