<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="pamay-arslan-etal-2026-corefinst">
    <titleInfo>
        <title>CorefInst: Leveraging LLMs for Multilingual Coreference Resolution</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Tuğba</namePart>
        <namePart type="family">Pamay Arslan</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Emircan</namePart>
        <namePart type="family">Erol</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Gülşen</namePart>
        <namePart type="family">Eryiğit</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <originInfo>
        <dateIssued>2026</dateIssued>
    </originInfo>
    <typeOfResource>text</typeOfResource>
    <genre authority="bibutilsgt">journal article</genre>
    <relatedItem type="host">
        <titleInfo>
            <title>Transactions of the Association for Computational Linguistics</title>
        </titleInfo>
        <originInfo>
            <issuance>continuing</issuance>
            <publisher>MIT Press</publisher>
            <place>
                <placeTerm type="text">Cambridge, MA</placeTerm>
            </place>
        </originInfo>
        <genre authority="marcgt">periodical</genre>
        <genre authority="bibutilsgt">academic journal</genre>
    </relatedItem>
    <abstract>Coreference Resolution (CR) is a crucial yet challenging task in natural language understanding, often constrained by task-specific architectures and encoder-based language models that demand extensive training and lack adaptability. This study introduces the first multilingual CR methodology which leverages decoder-only LLMs to handle both overt and zero mentions. The article explores how to model the CR task for LLMs via five different instruction sets using a controlled inference method. The approach is evaluated across three LLMs: Llama 3.1, Gemma 2, and Mistral 0.3. The results indicate that LLMs, when instruction-tuned with a suitable instruction set, can surpass state-of-the-art task-specific architectures. Specifically, our best model, a fully fine-tuned Llama 3.1 for multilingual CR, outperforms the leading multilingual CR model (i.e., Corpipe 24 single stage variant) by 2 percentage points on average across all languages in the CorefUD v1.2 dataset collection.</abstract>
    <identifier type="citekey">pamay-arslan-etal-2026-corefinst</identifier>
    <identifier type="doi">10.1162/tacl.a.593</identifier>
    <location>
        <url>https://aclanthology.org/2026.tacl-1.4/</url>
    </location>
    <part>
        <date>2026</date>
        <detail type="volume"><number>14</number></detail>
        <extent unit="page">
            <start>64</start>
            <end>80</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Journal Article
%T CorefInst: Leveraging LLMs for Multilingual Coreference Resolution
%A Pamay Arslan, Tuğba
%A Erol, Emircan
%A Eryiğit, Gülşen
%J Transactions of the Association for Computational Linguistics
%D 2026
%V 14
%I MIT Press
%C Cambridge, MA
%F pamay-arslan-etal-2026-corefinst
%X Coreference Resolution (CR) is a crucial yet challenging task in natural language understanding, often constrained by task-specific architectures and encoder-based language models that demand extensive training and lack adaptability. This study introduces the first multilingual CR methodology which leverages decoder-only LLMs to handle both overt and zero mentions. The article explores how to model the CR task for LLMs via five different instruction sets using a controlled inference method. The approach is evaluated across three LLMs: Llama 3.1, Gemma 2, and Mistral 0.3. The results indicate that LLMs, when instruction-tuned with a suitable instruction set, can surpass state-of-the-art task-specific architectures. Specifically, our best model, a fully fine-tuned Llama 3.1 for multilingual CR, outperforms the leading multilingual CR model (i.e., Corpipe 24 single stage variant) by 2 percentage points on average across all languages in the CorefUD v1.2 dataset collection.
%R 10.1162/tacl.a.593
%U https://aclanthology.org/2026.tacl-1.4/
%U https://doi.org/10.1162/tacl.a.593
%P 64-80

Markdown (Informal)

[CorefInst: Leveraging LLMs for Multilingual Coreference Resolution](https://aclanthology.org/2026.tacl-1.4/) (Pamay Arslan et al., TACL 2026)

• CorefInst: Leveraging LLMs for Multilingual Coreference Resolution (Pamay Arslan et al., TACL 2026)

ACL

• Tuğba Pamay Arslan, Emircan Erol, and Gülşen Eryiğit. 2026. CorefInst: Leveraging LLMs for Multilingual Coreference Resolution. Transactions of the Association for Computational Linguistics, 14:64–80.