Abstract:
The recent proliferation of social media has undoubtedly brought about many benefits, but along with it also came a serious impediment to society in the form of “fake news” which has become an eminent barrier to journalism, freedom of expression, and democracy as a whole. The study aims to understand the currently used AI techniques for detecting fake news, identify their shortcomings, and compare them with the emerging models. We compared the performance of memory-based methods (LSTM and Bi-LSTM) with traditional methods. We also compared the changes in performance after applying ensemble learning approaches. The study aimed to identify suitable models for fake news detection. This is in hopes of eventually promoting a safe and healthy environment for sharing information and content online and, in the process, helping develop strategies and techniques to curb the spread of fake news on social media.