Fake News Detection in Social Media using a Novel FakeBERT Approach

Abstract

Today, the news media has changed from offline to online, and this transformation will help the public to get information quickly and efficiently; in the same way, this media will spread phoney details rapidly. In recent research, many valuable methods were used to detect counterfeit information and analyse it unidirectional. In this research, we used bidirectional training approaches. We proposed two methods in this research. The first method is the deep learning approach as a Bidirectional Encoder Representation from Transformers (FakeBert) and a combination of Convolutional Neural Networks. This combination will help us manage the quality of detecting fake news. The proposed classification model FakeBert will provide better performance when compared to the existing model, and the accuracy is 99.90%.