Hate and Sarcasm Classification Using Machine Learning and Deep Learning Techniques: A Survey

Abstract

This survey paper confers a study on hateful and sarcasm classification techniques. The importance of hateful and sarcasm classification in social media is due to, the amount of online hate speech, and sarcasm speech growing day by day, and it affects human emotions. For this reason, hate and sarcasm speech has become a real problem in modern society and this needs to be controlled by using classification techniques to detect hate and sarcasm. The proposed survey describes the comparative study on datasets available, data pre-processing techniques used, and methods used for classification, which is used to automatically recognize hateful and sarcasm detection using both machine learning and deep learning techniques. The findings mainly addressed the Sentiment Based Features, Semantic-Based Features, N-grams, and Pattern-Based Features, which are most frequently and preferably used by the researchers in this field. Also, relative analysis on different machine learning and deep learning methods used for hate and sarcasm detection is analyzed in this work which helps the researchers in this field to choose better models for the specific data sets used.