Evaluating the Effectiveness of Hybrid Features in Fake News Detection on Social Media

This research focuses on the detection of fake news on social media, specifically in Amharic language posts. The study highlights the importance of utilizing hybrid features, which include both news content and social content features, to improve the accuracy of fake news detection. We evaluated the effectiveness of these hybrid features using state-of-the-art methodologies and explored methods that optimize detection accuracy and reduce latency. Our research revealed that we achieved an impressive Fl-score of 0.99 by utilizing a BERT-based uncased model. This outcome was obtained by incorporating a combination of textual content, publication dates, and page creation dates as hybrid features.