@inproceedings{yang-etal-2026-egr,
title = "{EGR}-{FV}: Evidence-Grounded Remixing for Debiased Fact Verification via Counterfactual Necessity-Guided Training",
author = "Yang, Jun and
Song, Dandan and
Wu, Zhijing and
Tian, Yuhang and
Ma, Huipeng and
Li, Chenhao and
Jin, Yizhou and
Feng, Guangyuan and
Li, Xudong",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.emnlp-main.813/",
pages = "18931--18945",
ISBN = "979-8-89176-459-0",
abstract = "Automatic fact verification (FV) predicts the veracity of a claim against retrieved evidence, yet modern verifiers often behave as shortcut-driven verifiers, predicting from surface regularities in the claim rather than the factual content of the evidence. Existing debiasing methods over-correct this by uniformly enforcing evidence reliance on every sample, thereby suppressing the valid information already carried by the claim itself. We propose Evidence-Grounded Remixing for Fact Verification (EGR-FV), which estimates a per-sample evidence necessity score by contrasting claim-with-evidence and claim-only predictions, and uses it to remix training batches so that debiasing is adapted to each sample{'}s actual reliance on evidence. On FEVER, PolitiHop, and HOVER together with their symmetric counterparts, EGR-FV achieves state-of-the-art accuracy and robustness, validating the effectiveness of EGR-FV."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="yang-etal-2026-egr">
<titleInfo>
<title>EGR-FV: Evidence-Grounded Remixing for Debiased Fact Verification via Counterfactual Necessity-Guided Training</title>
</titleInfo>
<name type="personal">
<namePart type="given">Jun</namePart>
<namePart type="family">Yang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Dandan</namePart>
<namePart type="family">Song</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zhijing</namePart>
<namePart type="family">Wu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yuhang</namePart>
<namePart type="family">Tian</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Huipeng</namePart>
<namePart type="family">Ma</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Chenhao</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yizhou</namePart>
<namePart type="family">Jin</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Guangyuan</namePart>
<namePart type="family">Feng</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xudong</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-10</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing</title>
</titleInfo>
<name type="personal">
<namePart type="given">Hung-Yi</namePart>
<namePart type="family">Lee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sunipa</namePart>
<namePart type="family">Dev</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jessy</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Desmond</namePart>
<namePart type="family">Elliott</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Budapest, Hungary</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-459-0</identifier>
</relatedItem>
<abstract>Automatic fact verification (FV) predicts the veracity of a claim against retrieved evidence, yet modern verifiers often behave as shortcut-driven verifiers, predicting from surface regularities in the claim rather than the factual content of the evidence. Existing debiasing methods over-correct this by uniformly enforcing evidence reliance on every sample, thereby suppressing the valid information already carried by the claim itself. We propose Evidence-Grounded Remixing for Fact Verification (EGR-FV), which estimates a per-sample evidence necessity score by contrasting claim-with-evidence and claim-only predictions, and uses it to remix training batches so that debiasing is adapted to each sample’s actual reliance on evidence. On FEVER, PolitiHop, and HOVER together with their symmetric counterparts, EGR-FV achieves state-of-the-art accuracy and robustness, validating the effectiveness of EGR-FV.</abstract>
<identifier type="citekey">yang-etal-2026-egr</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-main.813/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>18931</start>
<end>18945</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T EGR-FV: Evidence-Grounded Remixing for Debiased Fact Verification via Counterfactual Necessity-Guided Training
%A Yang, Jun
%A Song, Dandan
%A Wu, Zhijing
%A Tian, Yuhang
%A Ma, Huipeng
%A Li, Chenhao
%A Jin, Yizhou
%A Feng, Guangyuan
%A Li, Xudong
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-459-0
%F yang-etal-2026-egr
%X Automatic fact verification (FV) predicts the veracity of a claim against retrieved evidence, yet modern verifiers often behave as shortcut-driven verifiers, predicting from surface regularities in the claim rather than the factual content of the evidence. Existing debiasing methods over-correct this by uniformly enforcing evidence reliance on every sample, thereby suppressing the valid information already carried by the claim itself. We propose Evidence-Grounded Remixing for Fact Verification (EGR-FV), which estimates a per-sample evidence necessity score by contrasting claim-with-evidence and claim-only predictions, and uses it to remix training batches so that debiasing is adapted to each sample’s actual reliance on evidence. On FEVER, PolitiHop, and HOVER together with their symmetric counterparts, EGR-FV achieves state-of-the-art accuracy and robustness, validating the effectiveness of EGR-FV.
%U https://aclanthology.org/2026.emnlp-main.813/
%P 18931-18945
Markdown (Informal)
[EGR-FV: Evidence-Grounded Remixing for Debiased Fact Verification via Counterfactual Necessity-Guided Training](https://aclanthology.org/2026.emnlp-main.813/) (Yang et al., EMNLP 2026)
ACL
- Jun Yang, Dandan Song, Zhijing Wu, Yuhang Tian, Huipeng Ma, Chenhao Li, Yizhou Jin, Guangyuan Feng, and Xudong Li. 2026. EGR-FV: Evidence-Grounded Remixing for Debiased Fact Verification via Counterfactual Necessity-Guided Training. In Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing, pages 18931–18945, Budapest, Hungary. Association for Computational Linguistics.