Mc-DNN: Fake News Detection Using MultiChannel Deep Neural Networks

Abstract

With the advancement of technology, social media has become a major source of digital news due to its global exposure. This has led to an increase in spreading fake news and misinformation online. Humans cannot differentiate fake news from real news because they can be easily influenced. A lot of research work has been conducted for detecting fake news using artificial intelligence and machine learning. A large number of deep learning models and their architectural variants have been investigated, and many websites are utilizing these models directly or indirectly to detect fake news. However, state-of-the-arts demonstrate the limited accuracy in distinguishing fake news from the original news. The authors propose a multi-channel deep learning model, namely Mc-DNN, leveraging and processing the news headlines and news articles along different channels for differentiating fake or real news. They achieve the highest accuracy of 99.23% on ISOT Fake News Dataset and 94.68% on Fake News Data for Mc-DNN. Thus, they highly recommend the use of Mc-DNN for fake news detection.