Researchers have proposed a new framework called InfoPDF to improve the accuracy of fake news detection. The framework uses a combination of real and synthetic data to learn effective representations of how information spreads online. Synthetic data is generated using large language models, but it can be unreliable. InfoPDF addresses this issue by modeling each synthetic propagation graph as a probabilistic latent distribution, allowing for reliability-aware fusion with real data. This approach has been shown to outperform existing methods on three real-world datasets. --- Why it matters: This matters because fake news detection is a critical problem in AI research, and improving its accuracy can have significant social implications. InfoPDF's ability to learn effective representations of propagation data can help researchers develop more robust models for detecting and mitigating the spread of misinformation. Source: https://arxiv.org/abs/2605.02259