Ensemble Learning Approach to Fake News Detection Problem

This article is devoted to developing a method for fake news detection. Due to the immense consequences of fake news spreading, this problem has recently gained importance and popularity. However, the currently available methods’ effectiveness still leaves much room for improvement - a detailed analysis of recent literature allowed for identifying current trends. Thus, in this paper, we propose a novel approach to the given task based on ensemble learning methods. Such an approach allows combining several different strategies to detect fake news detections into a unified, cohesive methodology. The performed experiments that used the widely known “Liar, Liar Pants on Fire” benchmark dataset yielded promising results which surpassed outcomes described in the literature.