Fake News Detection on Social Media for Sustainable Trust-based Social Networking

Abstract

In today's society, social media plays a remarkable part in imparting information and news about various social events. Without consideration for the authenticity of these news, some fake news is disseminated through social networking websites or platforms and reaches millions of individuals. This dissemination has a huge impact on human social lives as well as in the political realm. Therefore, detecting fake news is a rapidly evolving area of research to empower trust-based social networking. In this research study, we proposed a stacked bi-directional LSTM based model for discriminating among fake and genuine news. In the initial step, we acquire the data of news descriptions which is forwarded to the preprocessing stage followed by encoding the words or text into a dense vector representation. Subsequently, the stacked bidirectional LSTM is employed to categorize the news articles into real or Fake. The suggested approach is validated using the Fake News Detection database, which is a freely accessible benchmark database. Furthermore, we have performed a different set of experiments with various weight optimizers. It is observed that the suggested technique shows encouraging outcomes in classifying the news as real or fake with 95% accuracy. © 2021 IEEE.