arXiv cs.AI· Paul He, Elke Kirschbaum, Shiva Kasiviswanathan·· 6 天前AI 评分29
QXR：面向自然语言主张集合的全局一致性修复框架
Global Consistency Repair for Natural-Language Claim Sets
AI 导读
针对主张集合可能局部合理、两两兼容却整体不一致的失效模式，研究者提出 QXR 自适应诊断与修复框架，将 LLM 视为带噪的集合级一致性判断器，定位最小冲突子集并通过删除小型命中集来修复主张集合。在 VitaminC、FEVER 及冲突证据 RAG 衍生基准上，QXR 比一次性直接判断保留更多有效主张且保持高精度，该工作已被 EMNLP 2026 主会接收。
正文 · 原文
Abstract:NLP systems increasingly aggregate claims from retrieved passages, documents, and generated reasoning steps. Yet current factuality methods mostly verify claims individually or compare pairs of claims, leaving open the question of whether the entire aggregated claim set can be retained coherently. This misses a fundamental failure mode: claims may be locally plausible and pairwise compatible while still being jointly inconsistent. We formalize global consistency repair: given an inconsistent natural-language claim set, preserve the largest coherent subset while localizing the compact sources of inconsistency. We introduce QXR, an adaptive diagnosis-and-repair framework that treats an LLM as a noisy set-level consistency judge, localizes minimal conflicting subsets, and repairs the claim set by deleting a small hitting set of claims. Across LLM judges and benchmarks derived from VitaminC, FEVER, and conflicting-evidence RAG, QXR preserves substantially more valid claims than direct all-at-once judging while maintaining high precision. These results suggest that factuality in claim aggregation systems should be treated as a set-level repair problem, not only as local verification.
| Comments: | Accepted to EMNLP 2026 (Main Conference) |
| Subjects: | Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2601.13600 [cs.AI] |
| (or arXiv:2601.13600v2 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2601.13600 arXiv-issued DOI via DataCite |
Submission history
From: Paul He [view email]
[v1]
Tue, 20 Jan 2026 05:02:35 UTC (82 KB)
[v2]
Fri, 9 Oct 2026 17:18:50 UTC (164 KB)
来源：arXiv cs.AI · arxiv.org