HiPo: Detecting Fake News via Historical and Multi-Modal Analyses of Social Media Posts

In recent years, fake news has been a primary concern since it plays a significant role in influencing the political, economic, and social spheres. The scientific community has proposed several solutions to detect such fraudulent information. However, such solutions are unsuitable for social media posts since they cannot extract sufficient information from one-line textual and graphical content or are highly dependent on prior knowledge, which may be unavailable in the case of unprecedented events (e.g., breaking news). This paper tackles this issue by proposing HiPo, a novel multi-modal historical post-based fake news detection method. By combining the features extracted from the graphical and textual content, HiPo assesses the truthfulness of a social media post by building its historical context from prior off-label posts with high similarity, therefore achieving online detection without maintaining a context or knowledge database. We evaluate the performance of HiPo via an exhaustive set of experiments involving four real-world datasets. Our method achieves a detection accuracy higher than 84%, outperforming state-of-the-art methods in most experimental instances.