A System for Fake News Detection by using Supervised Learning Model for Social Media Contents

The evolution of ICTs has dramatically increased the number of people with internet access, which has altered the way the information is consumed. As a result, fake news has become one of the main concerns because it could destabilize governments and put them at risk for contemporary society. This can be seen in the example, consider the United States. Electoral campaign, where the term “fake news” became famous as a result of the hoaxes impact on the final outcome. This research work studies the possibility of using deep learning techniques to discriminate against counterfeit news on the Internet using only their text. To do so, three different neural network architectures are suggested, one on the basis of BERT, Google's modern linguistic model that achieves cutting-edge results. In this project are the applications for the detection of `fake news,' which is misleading news stories from reputable sources of the NLP (Natural Language Processing) methods. This approach has been implemented and examined in the form of a software system. Can you build a prototype that can differentiate between “real” and “fake” news? In this novel fake news detection approach by SVM achieved the accuracy of 92 % and Naive Bayes achieved the accuracy of 73%. In this study SVM can better than Naive Bayes classifier model in new prediction approach.