CAGE: Cognitive Attribution Graphs for Faithful Inline Citation Generation in Long-Form Question Answering
2026-07-27 • Computation and Language
Computation and Language
AI summaryⓘ
The authors address a problem in long-form question answering where citations are often linked to related topics but don’t fully support the claims made. They identify that it's hard for existing systems to clearly match each part of an answer to the right source, leading to unsupported claims. Their method, called CAGE, first creates a detailed map showing which parts of the answer are backed by which documents before writing the final answer. This approach improves how well the citations match the claims, as shown in tests on several datasets. Overall, the authors demonstrate that explicitly organizing evidence helps make answers more trustworthy.
long-form question answeringlarge language modelscitation generationevidence attributionCognitive Map Inductionsupport subgraphssemantic answer unitsstructured citation reasoningASQA datasetELI5 dataset
Authors
Zhichao Yan, Shizhao Li, Jiapu Wang, Haoran Luo, Qingang Zhang, Jiaoyan Chen, Ru Li, Jeff Z. Pan
Abstract
Long-form question answering increasingly relies on retrieved evidence to make LLM outputs verifiable, with inline citations tracing claims to source documents. However, existing systems often attach citations that are topically related but insufficient to support their claims. We identify attribution ambiguity as a structural challenge: end-to-end generation must implicitly resolve combinatorial claim--document assignments, obscuring evidential boundaries and increasing the risk of evidence-boundary overrun, where claims exceed cited support. To address this challenge, we propose CAGE (Cognitive Attribution Graphs for Citation Generation), a two-stage framework that introduces an explicit cognitive attribution map before answer generation. CAGE first trains a plug-and-play Cognitive Map Induction Model to construct answer-centered support subgraphs, aligning each semantic answer unit with supporting documents through explicit relations. A Structured Citation Reasoning Model then realizes these units as sentence-level claims with map-aligned citations. Experiments on ASQA, ELI5, and ExpertQA show that CAGE achieves state-of-the-art performance, demonstrating the effectiveness of attribution-space contraction and map-guided citation generation.