RAEEUCCI2023-Comparative study of Hate Speech detection using different Machine Learning Classifiers

Social media involves people getting closer to different parts of the world. The Creation of networks among different types and regions of the world consists of a clash of opinions and interests. This may lead to sharing of hate amongst people. Many social media platforms involve the concept of ‘Freedom to speech.’ But this also incorporates some restrictions as it can lead to huge problems of spreading hate messages. To have a peaceful environment around and prevent any kind of misinformation, there is a need of having a mechanism to automatically detect hate speech. So, this research paper solves this problem by using five different classifiers with a dataset with three classes. The classes are classified as offensive speech, hate speech, or offensive speech. The classifiers involved are Gaussian Naive Bayes, Random Forest, Support Vector Machine classifier, Logistic Regression, Decision Tree Classifier, and a TF IDF -based feature engineering technique. Maximum accuracy was achieved through RandomForest which was over 90%. Moreover, accuracy can be increased by using other feature engineering techniques and a combination of classifiers.