Enhancing Fake-News Detection with Node-Level Topological Features

Researchers have found that adding two simple graph metrics—degree centrality and local clustering coefficient—to existing AI models significantly improves fake news detection. This lightweight enhancement explicitly flags key network roles like hubs and community structures. On the Politifact dataset, the method boosted the model's macro F1 score from 0.7753 to 0.8344, a 7.6% improvement, without requiring complex architectural changes. The approach provides an interpretable and easily reproducible template for other information-diffusion tasks.