@inproceedings{baek-etal-2026-internalizing,
title = "Internalizing Negation-Gated Logical Rules into {LLM}s for Document-Level Relation Extraction",
author = "Baek, Hye-Yoon and
Ji, Sangjun and
Seo, Jimyeung and
Koo, Hae-Yoon and
Jin, Xiongnan and
Oh, Byungkook",
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.2463/",
doi = "10.18653/2026.findings-emnlp.2463",
pages = "52994--53011",
ISBN = "979-8-89176-462-0",
abstract = "Document-level relation extraction (DocRE) requires integrating scattered evidence to uncover challenging implicit relations. While recent LLM-based approaches leverage rich prior knowledge for relation prediction, they suffer from two limitations. Pipeline-based approaches cause irrecoverable candidate pruning, and the absence of explicit logical constraints leaves spurious over-prediction unsuppressed. To address this, we propose Gated Logical-rule INTernalization (GLINT), a novel DocRE framework that integrates an LLM with end-to-end inductive rule reasoning within a fully differentiable architecture. GLINT preserves per-pair signals across all entity pairs through a Fused Fact Tensor with Gumbel-softmax selection, and refines LLM predictions via an inductive rule reasoner with per-atom learnable negation gates. Extensive experiments on three benchmark datasets show that GLINT outperforms existing LLM-based baselines."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="baek-etal-2026-internalizing">
<titleInfo>
<title>Internalizing Negation-Gated Logical Rules into LLMs for Document-Level Relation Extraction</title>
</titleInfo>
<name type="personal">
<namePart type="given">Hye-Yoon</namePart>
<namePart type="family">Baek</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sangjun</namePart>
<namePart type="family">Ji</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jimyeung</namePart>
<namePart type="family">Seo</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Hae-Yoon</namePart>
<namePart type="family">Koo</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xiongnan</namePart>
<namePart type="family">Jin</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Byungkook</namePart>
<namePart type="family">Oh</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>Document-level relation extraction (DocRE) requires integrating scattered evidence to uncover challenging implicit relations. While recent LLM-based approaches leverage rich prior knowledge for relation prediction, they suffer from two limitations. Pipeline-based approaches cause irrecoverable candidate pruning, and the absence of explicit logical constraints leaves spurious over-prediction unsuppressed. To address this, we propose Gated Logical-rule INTernalization (GLINT), a novel DocRE framework that integrates an LLM with end-to-end inductive rule reasoning within a fully differentiable architecture. GLINT preserves per-pair signals across all entity pairs through a Fused Fact Tensor with Gumbel-softmax selection, and refines LLM predictions via an inductive rule reasoner with per-atom learnable negation gates. Extensive experiments on three benchmark datasets show that GLINT outperforms existing LLM-based baselines.</abstract>
<identifier type="citekey">baek-etal-2026-internalizing</identifier>
<identifier type="doi">10.18653/2026.findings-emnlp.2463</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.2463/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>52994</start>
<end>53011</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Internalizing Negation-Gated Logical Rules into LLMs for Document-Level Relation Extraction
%A Baek, Hye-Yoon
%A Ji, Sangjun
%A Seo, Jimyeung
%A Koo, Hae-Yoon
%A Jin, Xiongnan
%A Oh, Byungkook
%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 baek-etal-2026-internalizing
%X Document-level relation extraction (DocRE) requires integrating scattered evidence to uncover challenging implicit relations. While recent LLM-based approaches leverage rich prior knowledge for relation prediction, they suffer from two limitations. Pipeline-based approaches cause irrecoverable candidate pruning, and the absence of explicit logical constraints leaves spurious over-prediction unsuppressed. To address this, we propose Gated Logical-rule INTernalization (GLINT), a novel DocRE framework that integrates an LLM with end-to-end inductive rule reasoning within a fully differentiable architecture. GLINT preserves per-pair signals across all entity pairs through a Fused Fact Tensor with Gumbel-softmax selection, and refines LLM predictions via an inductive rule reasoner with per-atom learnable negation gates. Extensive experiments on three benchmark datasets show that GLINT outperforms existing LLM-based baselines.
%R 10.18653/2026.findings-emnlp.2463
%U https://aclanthology.org/2026.findings-emnlp.2463/
%U https://doi.org/10.18653/2026.findings-emnlp.2463
%P 52994-53011
Markdown (Informal)
[Internalizing Negation-Gated Logical Rules into LLMs for Document-Level Relation Extraction](https://aclanthology.org/2026.findings-emnlp.2463/) (Baek et al., Findings 2026)
ACL