Search papers, labs, and topics across Lattice.
This paper introduces G-Defense, a graph-enhanced framework for explainable fake news detection that decomposes news claims into sub-claims, models their dependencies in a graph, and uses RAG to retrieve evidence and generate competing explanations for each sub-claim. A defense-like inference module assesses overall veracity based on the graph, and an LLM generates an intuitive explanation graph. Experiments show G-Defense achieves state-of-the-art performance in both veracity detection and explanation quality compared to existing methods.
LLMs can generate high-quality, explainable fake news detection even when relying solely on unverified reports, by decomposing claims and modeling dependencies in a graph.
Explainable fake news detection aims to assess the veracity of news claims while providing human-friendly explanations. Existing methods incorporating investigative journalism are often inefficient and struggle with breaking news. Recent advances in large language models (LLMs) enable leveraging externally retrieved reports as evidence for detection and explanation generation, but unverified reports may introduce inaccuracies. Moreover, effective explainable fake news detection should provide a comprehensible explanation for all aspects of a claim to assist the public in verifying its accuracy. To address these challenges, we propose a graph-enhanced defense framework (G-Defense) that provides fine-grained explanations based solely on unverified reports. Specifically, we construct a claim-centered graph by decomposing the news claim into several sub-claims and modeling their dependency relationships. For each sub-claim, we use the retrieval-augmented generation (RAG) technique to retrieve salient evidence and generate competing explanations. We then introduce a defense-like inference module based on the graph to assess the overall veracity. Finally, we prompt an LLM to generate an intuitive explanation graph. Experimental results demonstrate that G-Defense achieves state-of-the-art performance in both veracity detection and the quality of its explanations.
