Evidence-based Factual Error Correction
James Thorne, Andreas Vlachos
摘要
This paper introduces the task of factual error correction: performing edits to a claim so that the generated rewrite is better supported by evidence. This extends the well-studied task of fact verification by providing a mechanism to correct written texts that are refuted or only partially supported by evidence. We demonstrate that it is feasible to train factual error correction systems from existing fact checking datasets which only contain labeled claims accompanied by evidence, but not the correction. We achieve this by employing a two-stage distant supervision approach that incorporates evidence into masked claims when generating corrections. Our approach, based on the T5 transformer and using retrieved evidence, achieved better results than existing work which used a pointer copy network and gold evidence, producing accurate factual error corrections for 5x more instances in human evaluation and a .125 increase in SARI score. The evaluation is conducted on a dataset of 65,000 instances based on a recent fact verification shared task and we release it to enable further work on the task. 1 John Goodman had the lead role in The Babe. John Goodman had the lead role in # #. John Goodman had the lead role in The Babe.
问问这篇 Paper
智能体会读完全文。
Lune 把这篇 Paper 索引到了最后一个公式,引用它的顶会 Paper 也一样。你提问,回答直接引用原文。
引用它的顶会 Paper14
- RARR: Researching and Revising What Language Models Say, Using Language ModelsLuyu Gao, Zhuyun Dai, Panupong Pasupat, Anthony Chen 等ACL 2023 · 被引用 90 次
- FaVIQ: FAct Verification from Information-seeking QuestionsJungsoo Park, Sewon Min, Jaewoo Kang, Luke Zettlemoyer 等ACL 2022 · 被引用 47 次
- LOREN: Logic-Regularized Reasoning for Interpretable Fact VerificationJiangjie Chen, Qiaoben Bao, Changzhi Sun, Xinbo Zhang 等AAAI 2022 · 被引用 44 次
- S^3cMath: Spontaneous Step-Level Self-Correction Makes Large Language Models Better Mathematical ReasonersYuchen Yan, Jin Jiang, Yang Liu, Yixin Cao 等AAAI 2025 · 被引用 19 次
- Zero-shot Faithful Factual Error CorrectionKung-Hsiang Huang, Hou Pong Chan, Heng JiACL 2023 · 被引用 17 次
它引用的顶会 Paper7
- The Curious Case of Neural Text DegenerationAri Holtzman, Jan Buys, Li Du, Maxwell Forbes 等ICLR 2020 · 被引用 4,112 次
- Autoregressive Entity RetrievalNicola De Cao, Gautier Izacard, Sebastian Riedel, Fabio PetroniICLR 2021 · 被引用 200 次
- Dense Passage Retrieval for Open-Domain Question AnsweringVladimir Karpukhin, Barlas Oguz, Sewon Min, Patrick Lewis 等EMNLP 2020 · 被引用 142 次
- Generating Fact Checking ExplanationsPepa Atanasova, Jakob Grue Simonsen, Christina Lioma, Isabelle AugensteinACL 2020 · 被引用 130 次
- Automatic Fact-Guided Sentence ModificationDarsh J. Shah, Tal Schuster, Regina BarzilayAAAI 2020 · 被引用 44 次
相关 Paper
- Converge to the Truth: Factual Error Correction via Iterative Constrained EditingJiangjie Chen, Rui Xu, Wenxuan Zeng, Changzhi Sun 等AAAI 2023 · 被引用 13 次
- Improving Factual Error Correction by Learning to Inject Factual ErrorsXingwei He, Qianru Zhang, A-Long Jin, Jun Ma 等AAAI 2024 · 被引用 5 次
- LM vs LM: Detecting Factual Errors via Cross ExaminationRoi Cohen, May Hamri, Mor Geva, Amir GlobersonEMNLP 2023 · 被引用 41 次
- Mask-to-Correct⁺: Leveraging Retriever Diversity for Masking-guided Faithful Fact CorrectionPayel Santra, Lavisha Sharma, Madhusudan Ghosh, Partha BasuchowdhuriACL 2026
- DialFact: A Benchmark for Fact-Checking in DialoguePrakhar Gupta, Chien-Sheng Wu, Wenhao Liu, Caiming XiongACL 2022
