Abusive Language Detection using ML

Abstract

Detection of abusive language in online content has become a serious problem in recent years. Abusive language is an expression that carries grimy phrases each withinside the context of jokes, threat, vulgar intercourse conservation, or to disrespect someone. Nowadays many people post toxic comments in the social media such as Facebook, Twitter, etc. Many researches have tried to do it using the regular expression and blacklists but this is not the permanent solution for this problem that is why, we are using the machine learning to handle the all kind of comments written in English language. Abusive language detection is a hard trouble to clear up due to the fact this trouble cannot be resolved through phrase matching. This paper discusses a preliminaries study for toxic comment detection in English language and challenge in developing a model for English language comments used on Twitter, Facebook, etc. Because this is a classification problem, that is, we have used many machine learning algorithms like Naive Bayes, KNN, Random Forest, Decision Tree, and Logistic Algorithm for the prediction that the text is abusive or not abusive. After applying all the algorithms, we got to know that logistic algorithm was giving us the best results among all the algorithms.