A Supervised Machine Learning Approach to Fake News Identification

Abstract

In the era of digitization, we no longer need to struggle to connect to the outer world. There are so many sources that provide us with news reports of our surroundings, national and international activities. But the problem is there are many sources which are fooling people by spreading fake news about everything. This can be fatal sometimes as it encourages human rage. So to prevent this, the data scientists are eager to detect this kind of sources automatically. We have proposed a new model for this supervised data to predict if the provided news is fake or not. We have used some machine learning algorithms like- Gradient Boosting, Random Forest, Extra tree, XGBoost etc. and among these, GBM gives best result as accuracy of 95%. Then a majority voting classifier is made with these algorithm which gives accuracy of 94.15%.