A Classification method of Fake News based on Ensemble Learning

With1 the advent of the next generation of computing, news is available in various environments anytime, anywhere. This is a positive aspect of rapid information sharing, but information with unclear sources was produced in a news format and quickly spread to the public through social network services. The concept of fake news, which began to draw attention as of the 2016 U.S. presidential election, is now causing many economic and social damage around the world. As a result, IT and the industry are paying attention to classifying fake news and active research is ongoing. Therefore, identifying fake news and obtaining accurate information is a very important area in the information age. In this paper, after analyzing the Fake News Dataset of the ISOT, an Information Security and Object Technology, two methods of weighting were used. Based on this, Soft Voting Classifier, an ensemble method that showed the highest performance when using TF-IDF values as weight, is proposed as a fake news classification model.