Graph Alignment Topology as an Inductive Bias for Grounding Detection

In three linesMethod using aligned bipartite graphs and graph neural networks to detect hallucinations in LLMs. Trains a GNN on alignment structure between source documents and model outputs. SOTA results on 4 hallucination and QA datasets, outperforming GPT-4o.

Summary generated by Claude — human-verified