Natural Language Processing based Hybrid Model for Detecting Fake News Using Content-Based Features and Social Features

Internet acts as the best medium for proliferation and diffusion of fake news. Information quality on the internet is a very important issue, but web-scale data hinders the expert’s ability to correct much of the inaccurate content or fake content present over these platforms. Thus, a new system of safeguard is needed. Traditional Fake news detection systems are based on content-based features (i.e. analyzing the content of the news) of the news whereas most recent models focus on the social features of news (i.e. how the news is diffused in the network). This paper aims to build a novel machine learning model based on Natural Language Processing (NLP) techniques for the detection of ‘fake news’ by using both content-based features and social features of news. The proposed model has shown remarkable results and has achieved an average accuracy of 90.62% with F1 Score of 90.33% on a standard dataset.