@inproceedings{zhang-etal-2025-rethinking,
title = "Rethinking the Role of {LLM}s for Document-level Relation Extraction: a Refiner with Task Distribution and Probability Fusion",
author = "Zhang, Fu and
Jin, Xinlong and
Cheng, Jingwei and
Yu, Hongsen and
Xu, Huangming",
editor = "Chiruzzo, Luis and
Ritter, Alan and
Wang, Lu",
booktitle = "Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers)",
month = apr,
year = "2025",
address = "Albuquerque, New Mexico",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2025.naacl-long.319/",
doi = "10.18653/v1/2025.naacl-long.319",
pages = "6293--6312",
ISBN = "979-8-89176-189-6",
abstract = "Document-level relation extraction (DocRE) provides a broad context for extracting one or more relations for each entity pair. Large language models (LLMs) have made great progress in relation extraction tasks. However, one of the main challenges we face is that LLMs have difficulty in multi-label relation prediction tasks. Additionally, another noteworthy challenge and discovery we reveal: the small language models (SLMs) for DocRE tend to classify existing relations as ``no relation'' (NA), while LLMs tend to predict existing relations for all entity pairs. To address these challenges, we propose a novel method that utilizes LLMs as a refiner, employing task distribution and probability fusion. The task distribution we carefully designed aims to distinguish hard and easy tasks, and feed hard tasks to our LLMs-based framework to reevaluate and refine. Further, in order to effectively solve the multi-label relation prediction problem in the refinement process, we propose a probability fusion method, ensuring and enhancing fusion predictions by maintaining a balance between SLMs and LLMs. Extensive experiments on widely-used datasets demonstrate that our method outperforms existing LLMbased methods without fine-tuning by an average of 25.2{\%} F1. Refining SLMs using our method consistently boosts the performance of the SLMs, achieving new state-of-the-art results compared to existing SLMs and LLMs. Our code: https://github.com/Drasick/Drell."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="zhang-etal-2025-rethinking">
<titleInfo>
<title>Rethinking the Role of LLMs for Document-level Relation Extraction: a Refiner with Task Distribution and Probability Fusion</title>
</titleInfo>
<name type="personal">
<namePart type="given">Fu</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xinlong</namePart>
<namePart type="family">Jin</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jingwei</namePart>
<namePart type="family">Cheng</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Hongsen</namePart>
<namePart type="family">Yu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Huangming</namePart>
<namePart type="family">Xu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2025-04</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers)</title>
</titleInfo>
<name type="personal">
<namePart type="given">Luis</namePart>
<namePart type="family">Chiruzzo</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Alan</namePart>
<namePart type="family">Ritter</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Lu</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Albuquerque, New Mexico</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-189-6</identifier>
</relatedItem>
<abstract>Document-level relation extraction (DocRE) provides a broad context for extracting one or more relations for each entity pair. Large language models (LLMs) have made great progress in relation extraction tasks. However, one of the main challenges we face is that LLMs have difficulty in multi-label relation prediction tasks. Additionally, another noteworthy challenge and discovery we reveal: the small language models (SLMs) for DocRE tend to classify existing relations as “no relation” (NA), while LLMs tend to predict existing relations for all entity pairs. To address these challenges, we propose a novel method that utilizes LLMs as a refiner, employing task distribution and probability fusion. The task distribution we carefully designed aims to distinguish hard and easy tasks, and feed hard tasks to our LLMs-based framework to reevaluate and refine. Further, in order to effectively solve the multi-label relation prediction problem in the refinement process, we propose a probability fusion method, ensuring and enhancing fusion predictions by maintaining a balance between SLMs and LLMs. Extensive experiments on widely-used datasets demonstrate that our method outperforms existing LLMbased methods without fine-tuning by an average of 25.2% F1. Refining SLMs using our method consistently boosts the performance of the SLMs, achieving new state-of-the-art results compared to existing SLMs and LLMs. Our code: https://github.com/Drasick/Drell.</abstract>
<identifier type="citekey">zhang-etal-2025-rethinking</identifier>
<identifier type="doi">10.18653/v1/2025.naacl-long.319</identifier>
<location>
<url>https://aclanthology.org/2025.naacl-long.319/</url>
</location>
<part>
<date>2025-04</date>
<extent unit="page">
<start>6293</start>
<end>6312</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Rethinking the Role of LLMs for Document-level Relation Extraction: a Refiner with Task Distribution and Probability Fusion
%A Zhang, Fu
%A Jin, Xinlong
%A Cheng, Jingwei
%A Yu, Hongsen
%A Xu, Huangming
%Y Chiruzzo, Luis
%Y Ritter, Alan
%Y Wang, Lu
%S Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers)
%D 2025
%8 April
%I Association for Computational Linguistics
%C Albuquerque, New Mexico
%@ 979-8-89176-189-6
%F zhang-etal-2025-rethinking
%X Document-level relation extraction (DocRE) provides a broad context for extracting one or more relations for each entity pair. Large language models (LLMs) have made great progress in relation extraction tasks. However, one of the main challenges we face is that LLMs have difficulty in multi-label relation prediction tasks. Additionally, another noteworthy challenge and discovery we reveal: the small language models (SLMs) for DocRE tend to classify existing relations as “no relation” (NA), while LLMs tend to predict existing relations for all entity pairs. To address these challenges, we propose a novel method that utilizes LLMs as a refiner, employing task distribution and probability fusion. The task distribution we carefully designed aims to distinguish hard and easy tasks, and feed hard tasks to our LLMs-based framework to reevaluate and refine. Further, in order to effectively solve the multi-label relation prediction problem in the refinement process, we propose a probability fusion method, ensuring and enhancing fusion predictions by maintaining a balance between SLMs and LLMs. Extensive experiments on widely-used datasets demonstrate that our method outperforms existing LLMbased methods without fine-tuning by an average of 25.2% F1. Refining SLMs using our method consistently boosts the performance of the SLMs, achieving new state-of-the-art results compared to existing SLMs and LLMs. Our code: https://github.com/Drasick/Drell.
%R 10.18653/v1/2025.naacl-long.319
%U https://aclanthology.org/2025.naacl-long.319/
%U https://doi.org/10.18653/v1/2025.naacl-long.319
%P 6293-6312
Markdown (Informal)
[Rethinking the Role of LLMs for Document-level Relation Extraction: a Refiner with Task Distribution and Probability Fusion](https://aclanthology.org/2025.naacl-long.319/) (Zhang et al., NAACL 2025)
ACL