Web Text Content Credibility Analysis using Max Voting and Stacking Ensemble Classifiers

Abstract

The social media has become a great medium for people around the world to openly express their thoughts and views. But for all its advantages, it has also paved way for many people and organizations to intentionally spread fake news and misinform others. And the rate at which fake news is being currently generated, it has become critical to create a reliable mechanism that can efficiently classify a real news from a fake one. This research paper analyses the different approaches, involving ensemble learning, that can be used to accomplish the same by using only text features of the news data. We observe that a combination of three optimal ML algorithms, clubbed by an advanced ensemble learning technique, can give results with an accuracy of more than ninety eight percent.