@inproceedings{dong-etal-2026-star,
title = "{STAR} : Sentence Translation Alignment Rate for Document-to-Document Machine Translation",
author = "Dong, Yichen and
Wang, Hao and
Li, Junhui and
Xu, Linlong and
Wang, Longyue and
Luo, Weihua",
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.699/",
doi = "10.18653/2026.findings-emnlp.699",
pages = "14667--14689",
ISBN = "979-8-89176-462-0",
abstract = "Large Language Models (LLMs) have enabled a shift from sentence-level to document-to-document (Doc2Doc) machine translation, promising improved global coherence. However, document-to-document generation in a single pass frequently suffers from structural misalignment, manifesting as sentence omissions or hallucinations that violate the core requirement of source-target correspondence. To address this, we introduce **S**entence **T**ranslation **A**lignment **R**ate (**STAR**), an auxiliary metric that explicitly quantifies sentence-level structural fidelity. Building on this, we propose **STAR**-masked **P**reference **O**ptimization (**StarPO**), a framework that ranks document-level hypotheses by structural quality and utilizes a dynamic alignment mask to focus optimization on misaligned segments. Experimental results across news and literary domains demonstrate that StarPO significantly enhances translation quality and structural integrity. Notably, StarPO allows compact models to surpass the performance of massive proprietary systems like GPT-4o while maintaining superior token efficiency. We will release our code and datasets."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="dong-etal-2026-star">
<titleInfo>
<title>STAR : Sentence Translation Alignment Rate for Document-to-Document Machine Translation</title>
</titleInfo>
<name type="personal">
<namePart type="given">Yichen</namePart>
<namePart type="family">Dong</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Hao</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Junhui</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Linlong</namePart>
<namePart type="family">Xu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Longyue</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Weihua</namePart>
<namePart type="family">Luo</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>Large Language Models (LLMs) have enabled a shift from sentence-level to document-to-document (Doc2Doc) machine translation, promising improved global coherence. However, document-to-document generation in a single pass frequently suffers from structural misalignment, manifesting as sentence omissions or hallucinations that violate the core requirement of source-target correspondence. To address this, we introduce **S**entence **T**ranslation **A**lignment **R**ate (**STAR**), an auxiliary metric that explicitly quantifies sentence-level structural fidelity. Building on this, we propose **STAR**-masked **P**reference **O**ptimization (**StarPO**), a framework that ranks document-level hypotheses by structural quality and utilizes a dynamic alignment mask to focus optimization on misaligned segments. Experimental results across news and literary domains demonstrate that StarPO significantly enhances translation quality and structural integrity. Notably, StarPO allows compact models to surpass the performance of massive proprietary systems like GPT-4o while maintaining superior token efficiency. We will release our code and datasets.</abstract>
<identifier type="citekey">dong-etal-2026-star</identifier>
<identifier type="doi">10.18653/2026.findings-emnlp.699</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.699/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>14667</start>
<end>14689</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T STAR : Sentence Translation Alignment Rate for Document-to-Document Machine Translation
%A Dong, Yichen
%A Wang, Hao
%A Li, Junhui
%A Xu, Linlong
%A Wang, Longyue
%A Luo, Weihua
%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 dong-etal-2026-star
%X Large Language Models (LLMs) have enabled a shift from sentence-level to document-to-document (Doc2Doc) machine translation, promising improved global coherence. However, document-to-document generation in a single pass frequently suffers from structural misalignment, manifesting as sentence omissions or hallucinations that violate the core requirement of source-target correspondence. To address this, we introduce **S**entence **T**ranslation **A**lignment **R**ate (**STAR**), an auxiliary metric that explicitly quantifies sentence-level structural fidelity. Building on this, we propose **STAR**-masked **P**reference **O**ptimization (**StarPO**), a framework that ranks document-level hypotheses by structural quality and utilizes a dynamic alignment mask to focus optimization on misaligned segments. Experimental results across news and literary domains demonstrate that StarPO significantly enhances translation quality and structural integrity. Notably, StarPO allows compact models to surpass the performance of massive proprietary systems like GPT-4o while maintaining superior token efficiency. We will release our code and datasets.
%R 10.18653/2026.findings-emnlp.699
%U https://aclanthology.org/2026.findings-emnlp.699/
%U https://doi.org/10.18653/2026.findings-emnlp.699
%P 14667-14689
Markdown (Informal)
[STAR : Sentence Translation Alignment Rate for Document-to-Document Machine Translation](https://aclanthology.org/2026.findings-emnlp.699/) (Dong et al., Findings 2026)
ACL