Abstract: Abusive language is an wordingthat accommodate abusive words which can be in the context of jokes, or toinvoking someone. Nowadays almost every user make use of an abusive language inthe social media platform such as Facebook, Linkedin, Instagram, Twitter, etc.It is one of the difficult task to identifying an abusive word in huge world ofsocial media because this problem cannot be determined by simply word matching.With fast growing of social networks and communication between people fromdifferent countries and different state of mind has become more direct, whichresults into using more and more cursing words between these people. Therefore,it arises the need of detecting such speech automatically and modify any datathat contains abusive language. In this project, we propose an approach todetect abusive words from reviews and classify the reviews as positive ornegative. Our approach is based on unigrams and patterns that are automaticallycollected from the training set. We use Random Forest Decision Tree classifierto identify the review whether the review is abusive or not.

Keywords:- Abusiveword, Random Forest, Sentiment analysis, Decision tree, Data Pre-processing.