ABSTRACT
Social media has recently grown to be one of the leading entertainment areas. The increased use of social media platforms by people all over the world has given rise to several fascinating NLP challenges. One of the major problems with NLP is detecting taboos in streaming platform chats and comments. In past works, taboos were identified on different datasets using a range of deep learning algorithms and feature engineering techniques to address this growing problem. Although many research have already been conducted to detect hate material, most of these have been conducted in a single context, no study has been done to compare the comparison to machine learning or deep learning models perform the best on common public data. A publicly released dataset with three different classes will be used in this study to compare the performance of two feature engineering techniques and seven deep learning algorithms. The testing findings showed the high accuracy and less time complexity.
