Ensemble Based High Performance Deep Learning Model for Fake News Detection

Abstract

Since the advent of social media platforms such as Facebook and Twitter, information has been disseminated at a velocity never before seen in human history. Many users are submitting statements that are not based on reality because of the current trend in social media usage, which has led to an explosion of user-generated material. Determining if a text contains deceptive or inaccurate information is a challenging task to automate. Even an expert in the field has to look at an article from several perspectives before making a judgement on its credibility. Using an ensemble machine learning approach, we provide a way for automatically categorizing news items. Our study delves into several textual elements to distinguish between factual and fraudulent material. We use these features to train an ensemble of machine learning algorithms, and then we test them on four real-world datasets to see how well they did. We propose an ensemble learner strategy, and the results show that it outperforms the individual learner approaches.