FactNLI: Dynamic and Automated Fact-Based Augmentation of NLI Benchmarks

Yihan Lu; Junzhe Zhang; Huixuan Zhang; Xiaojun Wan

FactNLI: Dynamic and Automated Fact-Based Augmentation of NLI Benchmarks

Yihan Lu, Junzhe Zhang, Huixuan Zhang, Xiaojun Wan

Correct Metadata for

Use this form to create a GitHub issue with structured data describing the correction. You will need a GitHub account. Once you create that issue, the correction will be reviewed by a staff member.

⚠️ Mobile Users: Submitting this form to create a new issue will only work with github.com, not the GitHub Mobile app.

Important: The Anthology treat PDFs as authoritative. Please use this form only to correct data that is out of line with the PDF. See our corrections guidelines if you need to change the PDF.

Title Adjust the title. Retain tags such as <fixed-case>.

Authors Adjust author names and order to match the PDF.

Abstract Correct abstract if needed. Retain XML formatting tags such as <tex-math>. You may use ... for bold, ... for italic, ... for underline, <sc>...</sc> for small-caps, <tt>...<tt> for typewriter text, <url>...</url> for URLs, <a href=...> for hyperlinks, and <par/> for paragraph breaks.

Verification against PDF Ensure that the new title/authors match the snapshot below. (If there is no snapshot or it is too small, consult the PDF.)

Authors concatenated from the text boxes above:

ALL author names match the snapshot above—including middle initials, hyphens, and accents.

Abstract

Natural Language Inference (NLI) is a core task for language understanding, yet existing NLI datasets are static and increasingly saturated, allowing current Large Language Models (LLMs) to perform well without fully revealing their capabilities and shortcomings. To address this problem, we propose a new data augmentation framework that automatically builds more challenging NLI datasets by iteratively fusing rich facts into the premise and hypothesis of an NLI instance. We use a strict fact filter to ensure that fused facts are non-contradictory and non-redundant. Applied to SNLI and MNLI, our augmentation substantially increases data length and complexity, and the performance of a range of LLMs on the augmented datasets drops significantly (up to 30%). Ablation experiments and human quality checks confirm the high quality of the generated data.

Anthology ID:: 2026.findings-emnlp.1234
Volume:: Findings of the Association for Computational Linguistics: EMNLP 2026
Month:: October
Year:: 2026
Address:: Budapest, Hungary
Editors:: Hung-Yi Lee, Sunipa Dev, Jessy Li, Desmond Elliott
Venue:: Findings
SIG:: SIGDAT
Publisher:: Association for Computational Linguistics
Note:
Pages:: 26107–26127
Language:
URL:: https://aclanthology.org/2026.findings-emnlp.1234/
DOI:
Bibkey:
Cite (ACL):: Yihan Lu, Junzhe Zhang, Huixuan Zhang, and Xiaojun Wan. 2026. FactNLI: Dynamic and Automated Fact-Based Augmentation of NLI Benchmarks. In Findings of the Association for Computational Linguistics: EMNLP 2026, pages 26107–26127, Budapest, Hungary. Association for Computational Linguistics.
Cite (Informal):: FactNLI: Dynamic and Automated Fact-Based Augmentation of NLI Benchmarks (Lu et al., Findings 2026)
Copy Citation:
PDF:: https://aclanthology.org/2026.findings-emnlp.1234.pdf

PDF Cite Search Fix data

Export citation

BibTeX
MODS XML
Endnote
Preformatted

@inproceedings{lu-etal-2026-factnli,
    title = "{F}act{NLI}: Dynamic and Automated Fact-Based Augmentation of {NLI} Benchmarks",
    author = "Lu, Yihan  and
      Zhang, Junzhe  and
      Zhang, Huixuan  and
      Wan, Xiaojun",
    editor = "Lee, Hung-Yi  and
      Dev, Sunipa  and
      Li, Jessy  and
      Elliott, Desmond",
    booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {EMNLP} 2026",
    month = oct,
    year = "2026",
    address = "Budapest, Hungary",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2026.findings-emnlp.1234/",
    pages = "26107--26127",
    ISBN = "979-8-89176-462-0",
    abstract = "Natural Language Inference (NLI) is a core task for language understanding, yet existing NLI datasets are static and increasingly saturated, allowing current Large Language Models (LLMs) to perform well without fully revealing their capabilities and shortcomings. To address this problem, we propose a new data augmentation framework that automatically builds more challenging NLI datasets by iteratively fusing rich facts into the premise and hypothesis of an NLI instance. We use a strict fact filter to ensure that fused facts are non-contradictory and non-redundant. Applied to SNLI and MNLI, our augmentation substantially increases data length and complexity, and the performance of a range of LLMs on the augmented datasets drops significantly (up to 30{\%}). Ablation experiments and human quality checks confirm the high quality of the generated data."
}

Download as File

<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="lu-etal-2026-factnli">
    <titleInfo>
        <title>FactNLI: Dynamic and Automated Fact-Based Augmentation of NLI Benchmarks</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Yihan</namePart>
        <namePart type="family">Lu</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Junzhe</namePart>
        <namePart type="family">Zhang</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Huixuan</namePart>
        <namePart type="family">Zhang</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Xiaojun</namePart>
        <namePart type="family">Wan</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>Findings of the Association for Computational Linguistics: EMNLP 2026</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">979-8-89176-462-0</identifier>
    </relatedItem>
    <abstract>Natural Language Inference (NLI) is a core task for language understanding, yet existing NLI datasets are static and increasingly saturated, allowing current Large Language Models (LLMs) to perform well without fully revealing their capabilities and shortcomings. To address this problem, we propose a new data augmentation framework that automatically builds more challenging NLI datasets by iteratively fusing rich facts into the premise and hypothesis of an NLI instance. We use a strict fact filter to ensure that fused facts are non-contradictory and non-redundant. Applied to SNLI and MNLI, our augmentation substantially increases data length and complexity, and the performance of a range of LLMs on the augmented datasets drops significantly (up to 30%). Ablation experiments and human quality checks confirm the high quality of the generated data.</abstract>
    <identifier type="citekey">lu-etal-2026-factnli</identifier>
    <location>
        <url>https://aclanthology.org/2026.findings-emnlp.1234/</url>
    </location>
    <part>
        <date>2026-10</date>
        <extent unit="page">
            <start>26107</start>
            <end>26127</end>
        </extent>
    </part>
</mods>
</modsCollection>

Download as File

%0 Conference Proceedings
%T FactNLI: Dynamic and Automated Fact-Based Augmentation of NLI Benchmarks
%A Lu, Yihan
%A Zhang, Junzhe
%A Zhang, Huixuan
%A Wan, Xiaojun
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Findings of the Association for Computational Linguistics: EMNLP 2026
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-462-0
%F lu-etal-2026-factnli
%X Natural Language Inference (NLI) is a core task for language understanding, yet existing NLI datasets are static and increasingly saturated, allowing current Large Language Models (LLMs) to perform well without fully revealing their capabilities and shortcomings. To address this problem, we propose a new data augmentation framework that automatically builds more challenging NLI datasets by iteratively fusing rich facts into the premise and hypothesis of an NLI instance. We use a strict fact filter to ensure that fused facts are non-contradictory and non-redundant. Applied to SNLI and MNLI, our augmentation substantially increases data length and complexity, and the performance of a range of LLMs on the augmented datasets drops significantly (up to 30%). Ablation experiments and human quality checks confirm the high quality of the generated data.
%U https://aclanthology.org/2026.findings-emnlp.1234/
%P 26107-26127

Download as File

Markdown (Informal)

[FactNLI: Dynamic and Automated Fact-Based Augmentation of NLI Benchmarks](https://aclanthology.org/2026.findings-emnlp.1234/) (Lu et al., Findings 2026)

FactNLI: Dynamic and Automated Fact-Based Augmentation of NLI Benchmarks (Lu et al., Findings 2026)

ACL

Yihan Lu, Junzhe Zhang, Huixuan Zhang, and Xiaojun Wan. 2026. FactNLI: Dynamic and Automated Fact-Based Augmentation of NLI Benchmarks. In Findings of the Association for Computational Linguistics: EMNLP 2026, pages 26107–26127, Budapest, Hungary. Association for Computational Linguistics.