@inproceedings{liu-etal-2026-beyondatoms,
title = "{B}eyond{A}toms: Benchmarking and Verifying Relational Factuality Beyond Atomic Facts",
author = "Liu, Zeyang and
Xue, Jingfeng and
Fu, Jiarun and
Du, Wenbiao and
Yang, Xiuqi and
Fu, Yifeng and
Chen, Junbao and
Wang, Yong",
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.289/",
pages = "6620--6637",
ISBN = "979-8-89176-459-0",
abstract = "As Large Language Models (LLMs) increasingly generate reasoning-driven outputs that weave isolated events into intricate logical topologies, factuality depends not only on isolated atomic facts but critically on the faithful relations connecting them. However, current verification paradigms prioritize individual facts and remain fundamentally blind to relational distortions. This oversight creates a dangerous false sense of security, where verifiers inadvertently endorse relationally corrupted claims simply because their constituent events are factual. Furthermore, the community lacks a systematic benchmark to evaluate this crucial relational layer. To address these critical vulnerabilities, we introduce \textsc{BeyondAtoms}, a novel benchmark for relational fact verification. Our evaluations reveal that generic verification pipelines often fail by misclassifying relational manipulations as supported. While explicitly guided reasoning models can resolve these structural errors, they incur prohibitive computational costs. To counter this pervasive relation blindness, we propose CARVE, a completely training-free verification framework. By decoupling structural dependencies through deterministic extraction and bi-directional relational alignment, CARVE achieves a favorable accuracy{--}efficiency trade-off: it remains highly competitive on relational and general fact verification while avoiding the heavy inference cost of reasoning-based LLMs."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="liu-etal-2026-beyondatoms">
<titleInfo>
<title>BeyondAtoms: Benchmarking and Verifying Relational Factuality Beyond Atomic Facts</title>
</titleInfo>
<name type="personal">
<namePart type="given">Zeyang</namePart>
<namePart type="family">Liu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jingfeng</namePart>
<namePart type="family">Xue</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jiarun</namePart>
<namePart type="family">Fu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Wenbiao</namePart>
<namePart type="family">Du</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xiuqi</namePart>
<namePart type="family">Yang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yifeng</namePart>
<namePart type="family">Fu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Junbao</namePart>
<namePart type="family">Chen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yong</namePart>
<namePart type="family">Wang</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>As Large Language Models (LLMs) increasingly generate reasoning-driven outputs that weave isolated events into intricate logical topologies, factuality depends not only on isolated atomic facts but critically on the faithful relations connecting them. However, current verification paradigms prioritize individual facts and remain fundamentally blind to relational distortions. This oversight creates a dangerous false sense of security, where verifiers inadvertently endorse relationally corrupted claims simply because their constituent events are factual. Furthermore, the community lacks a systematic benchmark to evaluate this crucial relational layer. To address these critical vulnerabilities, we introduce BeyondAtoms, a novel benchmark for relational fact verification. Our evaluations reveal that generic verification pipelines often fail by misclassifying relational manipulations as supported. While explicitly guided reasoning models can resolve these structural errors, they incur prohibitive computational costs. To counter this pervasive relation blindness, we propose CARVE, a completely training-free verification framework. By decoupling structural dependencies through deterministic extraction and bi-directional relational alignment, CARVE achieves a favorable accuracy–efficiency trade-off: it remains highly competitive on relational and general fact verification while avoiding the heavy inference cost of reasoning-based LLMs.</abstract>
<identifier type="citekey">liu-etal-2026-beyondatoms</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-main.289/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>6620</start>
<end>6637</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T BeyondAtoms: Benchmarking and Verifying Relational Factuality Beyond Atomic Facts
%A Liu, Zeyang
%A Xue, Jingfeng
%A Fu, Jiarun
%A Du, Wenbiao
%A Yang, Xiuqi
%A Fu, Yifeng
%A Chen, Junbao
%A Wang, Yong
%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 liu-etal-2026-beyondatoms
%X As Large Language Models (LLMs) increasingly generate reasoning-driven outputs that weave isolated events into intricate logical topologies, factuality depends not only on isolated atomic facts but critically on the faithful relations connecting them. However, current verification paradigms prioritize individual facts and remain fundamentally blind to relational distortions. This oversight creates a dangerous false sense of security, where verifiers inadvertently endorse relationally corrupted claims simply because their constituent events are factual. Furthermore, the community lacks a systematic benchmark to evaluate this crucial relational layer. To address these critical vulnerabilities, we introduce BeyondAtoms, a novel benchmark for relational fact verification. Our evaluations reveal that generic verification pipelines often fail by misclassifying relational manipulations as supported. While explicitly guided reasoning models can resolve these structural errors, they incur prohibitive computational costs. To counter this pervasive relation blindness, we propose CARVE, a completely training-free verification framework. By decoupling structural dependencies through deterministic extraction and bi-directional relational alignment, CARVE achieves a favorable accuracy–efficiency trade-off: it remains highly competitive on relational and general fact verification while avoiding the heavy inference cost of reasoning-based LLMs.
%U https://aclanthology.org/2026.emnlp-main.289/
%P 6620-6637
Markdown (Informal)
[BeyondAtoms: Benchmarking and Verifying Relational Factuality Beyond Atomic Facts](https://aclanthology.org/2026.emnlp-main.289/) (Liu et al., EMNLP 2026)
ACL
- Zeyang Liu, Jingfeng Xue, Jiarun Fu, Wenbiao Du, Xiuqi Yang, Yifeng Fu, Junbao Chen, and Yong Wang. 2026. BeyondAtoms: Benchmarking and Verifying Relational Factuality Beyond Atomic Facts. In Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing, pages 6620–6637, Budapest, Hungary. Association for Computational Linguistics.