STRA-MT: Deterministic Terminology Control for Document-Level Machine Translation

Vania Velda; Yigyu Hwang; Yongho Kim

STRA-MT: Deterministic Terminology Control for Document-Level Machine Translation

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

STRA-MT is a translation pipeline with deter- ministic terminology control built around small open-weight language models. The pipeline (1) reuses exact repetitions and guides near- duplicates using translations from earlier occur- rences, (2) maintains a growing terminology ledger initialized from either the provided glos- sary or a termbase mined from sample bitext, and (3) verifies every required term and, when one is missing, regenerates the paragraph once with the missing terms named. On the WMT25 zh-Hant→en development sets, STRA-MT im- proves mean terminology success over glossary prompting from 87.3 to 92.4 under the official WMT25 evaluation, while maintaining com- petitive BLEU, chrF++, and COMET scores. Official WMT26 test sets shows that term de- livery on en→pl comes close to the reference ceiling while the translation quality depends on the backbones.

Anthology ID:: 2026.wmt-main.123
Volume:: Proceedings of the Eleventh Conference on Machine Translation
Month:: October
Year:: 2026
Address:: Budapest, Hungary
Editors:: Barry Haddow, Tom Kocmi, Philipp Koehn, Christof Monz
Venues:: WMT | WS
SIG:: SIGMT
Publisher:: Association for Computational Linguistics
Note:
Pages:: 1881–1891
Language:
URL:: https://aclanthology.org/2026.wmt-main.123/
DOI:
Bibkey:
Cite (ACL):: Vania Velda, Yigyu Hwang, and Yongho Kim. 2026. STRA-MT: Deterministic Terminology Control for Document-Level Machine Translation. In Proceedings of the Eleventh Conference on Machine Translation, pages 1881–1891, Budapest, Hungary. Association for Computational Linguistics.
Cite (Informal):: STRA-MT: Deterministic Terminology Control for Document-Level Machine Translation (Velda et al., WMT 2026)
Copy Citation:
PDF:: https://aclanthology.org/2026.wmt-main.123.pdf

PDF Cite Search Fix data

Export citation

BibTeX
MODS XML
Endnote
Preformatted

@inproceedings{velda-etal-2026-stra,
    title = "{STRA}-{MT}: Deterministic Terminology Control for Document-Level Machine Translation",
    author = "Velda, Vania  and
      Hwang, Yigyu  and
      Kim, Yongho",
    editor = "Haddow, Barry  and
      Kocmi, Tom  and
      Koehn, Philipp  and
      Monz, Christof",
    booktitle = "Proceedings of the Eleventh Conference on Machine Translation",
    month = oct,
    year = "2026",
    address = "Budapest, Hungary",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2026.wmt-main.123/",
    pages = "1881--1891",
    ISBN = "979-8-89176-474-3",
    abstract = "STRA-MT is a translation pipeline with deter- ministic terminology control built around small open-weight language models. The pipeline (1) reuses exact repetitions and guides near- duplicates using translations from earlier occur- rences, (2) maintains a growing terminology ledger initialized from either the provided glos- sary or a termbase mined from sample bitext, and (3) verifies every required term and, when one is missing, regenerates the paragraph once with the missing terms named. On the WMT25 zh-Hant{\textrightarrow}en development sets, STRA-MT im- proves mean terminology success over glossary prompting from 87.3 to 92.4 under the official WMT25 evaluation, while maintaining com- petitive BLEU, chrF++, and COMET scores. Official WMT26 test sets shows that term de- livery on en{\textrightarrow}pl comes close to the reference ceiling while the translation quality depends on the backbones."
}

Download as File

<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="velda-etal-2026-stra">
    <titleInfo>
        <title>STRA-MT: Deterministic Terminology Control for Document-Level Machine Translation</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Vania</namePart>
        <namePart type="family">Velda</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Yigyu</namePart>
        <namePart type="family">Hwang</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Yongho</namePart>
        <namePart type="family">Kim</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>Proceedings of the Eleventh Conference on Machine Translation</title>
        </titleInfo>
        <name type="personal">
            <namePart type="given">Barry</namePart>
            <namePart type="family">Haddow</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Tom</namePart>
            <namePart type="family">Kocmi</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Philipp</namePart>
            <namePart type="family">Koehn</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Christof</namePart>
            <namePart type="family">Monz</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">979-8-89176-474-3</identifier>
    </relatedItem>
    <abstract>STRA-MT is a translation pipeline with deter- ministic terminology control built around small open-weight language models. The pipeline (1) reuses exact repetitions and guides near- duplicates using translations from earlier occur- rences, (2) maintains a growing terminology ledger initialized from either the provided glos- sary or a termbase mined from sample bitext, and (3) verifies every required term and, when one is missing, regenerates the paragraph once with the missing terms named. On the WMT25 zh-Hant→en development sets, STRA-MT im- proves mean terminology success over glossary prompting from 87.3 to 92.4 under the official WMT25 evaluation, while maintaining com- petitive BLEU, chrF++, and COMET scores. Official WMT26 test sets shows that term de- livery on en→pl comes close to the reference ceiling while the translation quality depends on the backbones.</abstract>
    <identifier type="citekey">velda-etal-2026-stra</identifier>
    <location>
        <url>https://aclanthology.org/2026.wmt-main.123/</url>
    </location>
    <part>
        <date>2026-10</date>
        <extent unit="page">
            <start>1881</start>
            <end>1891</end>
        </extent>
    </part>
</mods>
</modsCollection>

Download as File

%0 Conference Proceedings
%T STRA-MT: Deterministic Terminology Control for Document-Level Machine Translation
%A Velda, Vania
%A Hwang, Yigyu
%A Kim, Yongho
%Y Haddow, Barry
%Y Kocmi, Tom
%Y Koehn, Philipp
%Y Monz, Christof
%S Proceedings of the Eleventh Conference on Machine Translation
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-474-3
%F velda-etal-2026-stra
%X STRA-MT is a translation pipeline with deter- ministic terminology control built around small open-weight language models. The pipeline (1) reuses exact repetitions and guides near- duplicates using translations from earlier occur- rences, (2) maintains a growing terminology ledger initialized from either the provided glos- sary or a termbase mined from sample bitext, and (3) verifies every required term and, when one is missing, regenerates the paragraph once with the missing terms named. On the WMT25 zh-Hant→en development sets, STRA-MT im- proves mean terminology success over glossary prompting from 87.3 to 92.4 under the official WMT25 evaluation, while maintaining com- petitive BLEU, chrF++, and COMET scores. Official WMT26 test sets shows that term de- livery on en→pl comes close to the reference ceiling while the translation quality depends on the backbones.
%U https://aclanthology.org/2026.wmt-main.123/
%P 1881-1891

Download as File

Markdown (Informal)

[STRA-MT: Deterministic Terminology Control for Document-Level Machine Translation](https://aclanthology.org/2026.wmt-main.123/) (Velda et al., WMT 2026)

STRA-MT: Deterministic Terminology Control for Document-Level Machine Translation (Velda et al., WMT 2026)

ACL

Vania Velda, Yigyu Hwang, and Yongho Kim. 2026. STRA-MT: Deterministic Terminology Control for Document-Level Machine Translation. In Proceedings of the Eleventh Conference on Machine Translation, pages 1881–1891, Budapest, Hungary. Association for Computational Linguistics.