<?xml version="1.0" encoding="utf-8"?>
<TEI xmlns="http://www.tei-c.org/ns/1.0" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns:hal="http://hal.archives-ouvertes.fr/" xmlns:gml="http://www.opengis.net/gml/3.3/" xmlns:gmlce="http://www.opengis.net/gml/3.3/ce" version="1.1" xsi:schemaLocation="http://www.tei-c.org/ns/1.0 http://api.archives-ouvertes.fr/documents/aofr-sword.xsd">
  <teiHeader>
    <fileDesc>
      <titleStmt>
        <title>HAL TEI export of hal-05316675</title>
      </titleStmt>
      <publicationStmt>
        <distributor>CCSD</distributor>
        <availability status="restricted">
          <licence target="https://creativecommons.org/publicdomain/zero/1.0/">CC0 1.0 - Universal</licence>
        </availability>
        <date when="2026-05-22T21:58:10+02:00"/>
      </publicationStmt>
      <sourceDesc>
        <p part="N">HAL API Platform</p>
      </sourceDesc>
    </fileDesc>
  </teiHeader>
  <text>
    <body>
      <listBibl>
        <biblFull>
          <titleStmt>
            <title xml:lang="en">Improving Retrieval-Augmented Neural Machine Translation with Monolingual Data</title>
            <title xml:lang="fr">Améliorer la Traduction Neuronale par Exemple avec des Données Monolingues</title>
            <author role="aut">
              <persName>
                <forename type="first">Maxime</forename>
                <surname>Bouthors</surname>
              </persName>
              <email type="md5">88828f769c22db3829ab74ac844fa530</email>
              <email type="domain">isir.upmc.fr</email>
              <idno type="idhal" notation="string">maxime-bouthors</idno>
              <idno type="idhal" notation="numeric">1328527</idno>
              <idno type="halauthorid" notation="string">2983367-1328527</idno>
              <idno type="IDREF">https://www.idref.fr/285517902</idno>
              <idno type="ORCID">https://orcid.org/0000-0003-1400-4902</idno>
              <affiliation ref="#struct-1099426"/>
            </author>
            <author role="aut">
              <persName>
                <forename type="first">Josep</forename>
                <surname>Crego</surname>
              </persName>
              <email type="md5">7b71756e0526a2a84a6af09987ae45c0</email>
              <email type="domain">gps.tsc.upc.edu</email>
              <idno type="idhal" notation="numeric">1136018</idno>
              <idno type="halauthorid" notation="string">598689-1136018</idno>
              <idno type="ORCID">https://orcid.org/0009-0006-8034-3581</idno>
              <idno type="IDREF">https://www.idref.fr/259423823</idno>
              <affiliation ref="#struct-1099426"/>
            </author>
            <author role="aut">
              <persName>
                <forename type="first">François</forename>
                <surname>Yvon</surname>
              </persName>
              <email type="md5">c62025aca0c6cf8d65b4fe9600c991af</email>
              <email type="domain">cnrs.fr</email>
              <idno type="idhal" notation="string">francois-yvon</idno>
              <idno type="idhal" notation="numeric">5347</idno>
              <idno type="halauthorid" notation="string">2529017-5347</idno>
              <idno type="ORCID">https://orcid.org/0000-0002-7972-7442</idno>
              <idno type="IDREF">https://www.idref.fr/057593531</idno>
              <idno type="GOOGLE SCHOLAR">https://scholar.google.fr/citations?user=JjfDvawAAAAJ</idno>
              <idno type="ARXIV">https://arxiv.org/a/yvon_f_1</idno>
              <affiliation ref="#struct-1085231"/>
            </author>
            <editor role="depositor">
              <persName>
                <forename>François</forename>
                <surname>Yvon</surname>
              </persName>
              <email type="md5">c62025aca0c6cf8d65b4fe9600c991af</email>
              <email type="domain">cnrs.fr</email>
            </editor>
            <funder ref="#projanr-109431"/>
          </titleStmt>
          <editionStmt>
            <edition n="v1" type="current">
              <date type="whenSubmitted">2025-10-15 17:52:30</date>
              <date type="whenModified">2026-04-03 14:29:24</date>
              <date type="whenReleased">2025-10-16 08:01:22</date>
              <date type="whenProduced">2025-07-02</date>
              <date type="whenEndEmbargoed">2025-10-15</date>
              <ref type="file" target="https://hal.science/hal-05316675v1/document">
                <date notBefore="2025-10-15"/>
              </ref>
              <ref type="file" subtype="greenPublisher" n="1" target="https://hal.science/hal-05316675v1/file/2025.jeptalnrecital-taln.7.pdf" id="file-5316675-4554244">
                <date notBefore="2025-10-15"/>
              </ref>
            </edition>
            <respStmt>
              <resp>contributor</resp>
              <name key="164853">
                <persName>
                  <forename>François</forename>
                  <surname>Yvon</surname>
                </persName>
                <email type="md5">c62025aca0c6cf8d65b4fe9600c991af</email>
                <email type="domain">cnrs.fr</email>
              </name>
            </respStmt>
          </editionStmt>
          <publicationStmt>
            <distributor>CCSD</distributor>
            <idno type="halId">hal-05316675</idno>
            <idno type="halUri">https://hal.science/hal-05316675</idno>
            <idno type="halBibtex">bouthors:hal-05316675</idno>
            <idno type="halRefHtml">&lt;i&gt;32ème Conférence sur le Traitement Automatique des Langues Naturelles (TALN)&lt;/i&gt;, ATALA, Jul 2025, Marseille, France</idno>
            <idno type="halRef">32ème Conférence sur le Traitement Automatique des Langues Naturelles (TALN), ATALA, Jul 2025, Marseille, France</idno>
            <availability status="restricted">
              <licence target="https://creativecommons.org/licenses/by/4.0/">CC BY 4.0 - Attribution<ref corresp="#file-5316675-4554244"/></licence>
            </availability>
          </publicationStmt>
          <seriesStmt>
            <idno type="stamp" n="CNRS">CNRS - Centre national de la recherche scientifique</idno>
            <idno type="stamp" n="ISIR" corresp="SORBONNE-UNIVERSITE">Institut des Systèmes Intelligents et de Robotique</idno>
            <idno type="stamp" n="GENCI">Publications ayant eu recours aux supercalculateurs du GENCI</idno>
            <idno type="stamp" n="SORBONNE-UNIVERSITE">Sorbonne Université</idno>
            <idno type="stamp" n="SORBONNE-UNIV" corresp="SORBONNE-UNIVERSITE">Sorbonne Université 01/01/2018</idno>
            <idno type="stamp" n="SU-SCIENCES" corresp="SORBONNE-UNIVERSITE">Faculté des Sciences de Sorbonne Université</idno>
            <idno type="stamp" n="TALN-RECITAL">Conférences TALN RECITAL</idno>
            <idno type="stamp" n="SU-TI">Sorbonne Université - Texte Intégral</idno>
            <idno type="stamp" n="ANR">ANR</idno>
            <idno type="stamp" n="ALLIANCE-SU"> Alliance Sorbonne Université</idno>
            <idno type="stamp" n="ISIR_MLIA" corresp="ISIR">Machine Learning and Information Access</idno>
            <idno type="stamp" n="SUPRA_INGENIERIE">Ingénierie</idno>
            <idno type="stamp" n="ANR-IA-23" corresp="ANR-IA">ANR-IA-23</idno>
            <idno type="stamp" n="ANR-IA">ANR-IA</idno>
            <idno type="stamp" n="TEST-EMILIE-BUG">TEST-EMILIE-BUG</idno>
            <idno type="stamp" n="CORIA-TALN2025" corresp="TALN-RECITAL">Conférence CORIA TALN 2025</idno>
          </seriesStmt>
          <notesStmt>
            <note type="audience" n="2">International</note>
            <note type="invited" n="0">No</note>
            <note type="popular" n="0">No</note>
            <note type="peer" n="1">Yes</note>
            <note type="proceedings" n="1">Yes</note>
          </notesStmt>
          <sourceDesc>
            <biblStruct>
              <analytic>
                <title xml:lang="en">Improving Retrieval-Augmented Neural Machine Translation with Monolingual Data</title>
                <title xml:lang="fr">Améliorer la Traduction Neuronale par Exemple avec des Données Monolingues</title>
                <author role="aut">
                  <persName>
                    <forename type="first">Maxime</forename>
                    <surname>Bouthors</surname>
                  </persName>
                  <email type="md5">88828f769c22db3829ab74ac844fa530</email>
                  <email type="domain">isir.upmc.fr</email>
                  <idno type="idhal" notation="string">maxime-bouthors</idno>
                  <idno type="idhal" notation="numeric">1328527</idno>
                  <idno type="halauthorid" notation="string">2983367-1328527</idno>
                  <idno type="IDREF">https://www.idref.fr/285517902</idno>
                  <idno type="ORCID">https://orcid.org/0000-0003-1400-4902</idno>
                  <affiliation ref="#struct-1099426"/>
                </author>
                <author role="aut">
                  <persName>
                    <forename type="first">Josep</forename>
                    <surname>Crego</surname>
                  </persName>
                  <email type="md5">7b71756e0526a2a84a6af09987ae45c0</email>
                  <email type="domain">gps.tsc.upc.edu</email>
                  <idno type="idhal" notation="numeric">1136018</idno>
                  <idno type="halauthorid" notation="string">598689-1136018</idno>
                  <idno type="ORCID">https://orcid.org/0009-0006-8034-3581</idno>
                  <idno type="IDREF">https://www.idref.fr/259423823</idno>
                  <affiliation ref="#struct-1099426"/>
                </author>
                <author role="aut">
                  <persName>
                    <forename type="first">François</forename>
                    <surname>Yvon</surname>
                  </persName>
                  <email type="md5">c62025aca0c6cf8d65b4fe9600c991af</email>
                  <email type="domain">cnrs.fr</email>
                  <idno type="idhal" notation="string">francois-yvon</idno>
                  <idno type="idhal" notation="numeric">5347</idno>
                  <idno type="halauthorid" notation="string">2529017-5347</idno>
                  <idno type="ORCID">https://orcid.org/0000-0002-7972-7442</idno>
                  <idno type="IDREF">https://www.idref.fr/057593531</idno>
                  <idno type="GOOGLE SCHOLAR">https://scholar.google.fr/citations?user=JjfDvawAAAAJ</idno>
                  <idno type="ARXIV">https://arxiv.org/a/yvon_f_1</idno>
                  <affiliation ref="#struct-1085231"/>
                </author>
              </analytic>
              <monogr>
                <title level="m">Actes de la 32ème Conférence sur le Traitement Automatique des Langues Naturelles (TALN)</title>
                <meeting>
                  <title>32ème Conférence sur le Traitement Automatique des Langues Naturelles (TALN)</title>
                  <date type="start">2025-07-02</date>
                  <date type="end">2027-07-05</date>
                  <settlement>Marseille</settlement>
                  <country key="FR">France</country>
                </meeting>
                <respStmt>
                  <resp>conferenceOrganizer</resp>
                  <name>ATALA</name>
                </respStmt>
                <imprint/>
              </monogr>
              <ref type="publisher">https://aclanthology.org/2025.jeptalnrecital-taln.7/</ref>
            </biblStruct>
          </sourceDesc>
          <profileDesc>
            <langUsage>
              <language ident="fr">French</language>
            </langUsage>
            <textClass>
              <keywords scheme="author">
                <term xml:lang="en">neural machine translation</term>
                <term xml:lang="en">information retrieval</term>
                <term xml:lang="en">cross-lingual retrieval</term>
                <term xml:lang="en">retrieval augmented translation</term>
                <term xml:lang="fr">traduction neuronale</term>
                <term xml:lang="fr">recherche d’information</term>
                <term xml:lang="fr">recherche cross-lingue</term>
                <term xml:lang="fr">traduction à base d’exemples</term>
              </keywords>
              <classCode scheme="halDomain" n="info.info-tt">Computer Science [cs]/Document and Text Processing</classCode>
              <classCode scheme="halTypology" n="COMM">Conference papers</classCode>
              <classCode scheme="halOldTypology" n="COMM">Conference papers</classCode>
              <classCode scheme="halTreeTypology" n="COMM">Conference papers</classCode>
            </textClass>
            <abstract xml:lang="en">
              <p>Conventional retrieval-augmented neural machine translation (RANMT) systems leverage bilingual corpora, e.g., translation memories (TMs). Yet, in many settings, in-domain monolingual target-side corpora are often available. This work explores ways to take advantage of such resources by retrieving relevant segments directly in the target language, based on a source-side query. For this, we design improved cross-lingual retrieval systems, trained with both sentence level and word-level matching objectives. In our experiments with two RANMT architectures, we first demonstrate the benefits of such cross-lingual objectives in a controlled setting, obtaining translation performances that surpass standard TM-based models. We then showcase our method on a real-world set-up, where the target monolingual resources far exceed the amount of parallel data and observe large improvements of our new techniques, which outperform both the baseline setting, and general-purpose cross-lingual retrievers.</p>
            </abstract>
            <abstract xml:lang="fr">
              <p>Les systèmes de traduction neuronale augmentée par des exemples (RANMT) utilisent des corpus bilingues dits mémoires de traduction (TM). Pourtant, dans de nombreux cas, des corpus monolingues du domaine d'intérêt dans la langue cible sont disponibles. Nos travaux s'intéressent à l'exploitation de telles ressources, en recherchant les segments pertinents directement dans la langue cible, conditionnellement à une phrase source en requête. À cet effet, nous proposons d'améliorer les systèmes de recherche cross-lingue, en les entraînant à réaliser des association lexicales. Nos expériences avec deux architectures neuronales montrent l'avantage de notre méthode dans un cas contrôlé, conduisant à des performances de traduction qui peuvent surpasser les méthodes basées sur une mémoire de traduction. Enfin, nous évaluons notre méthode dans une configuration réaliste pour laquelle la quantité de données monolingues excède celle des données parallèles. Cette approche résulte en une nette amélioration des performances par rapport à des modèles de base ainsi que des encodeurs pré-entraînés.</p>
            </abstract>
            <particDesc>
              <org type="ir" xml:id="fr-esr-fdr-ir-064">
                <orgName type="acronym">GENCI</orgName>
                <idno type="wikidata">Q3093180</idno>
              </org>
            </particDesc>
          </profileDesc>
        </biblFull>
      </listBibl>
    </body>
    <back>
      <listOrg type="structures">
        <org type="institution" xml:id="struct-1099426" status="VALID">
          <orgName>ChapsVision</orgName>
          <desc>
            <address>
              <addrLine>4 rue du Port aux Vins, F-92 150 Suresnes</addrLine>
              <country key="FR"/>
            </address>
          </desc>
        </org>
        <org type="researchteam" xml:id="struct-1085231" status="VALID">
          <orgName>Machine Learning and Information Access</orgName>
          <orgName type="acronym">MLIA</orgName>
          <date type="start">2022-01-01</date>
          <desc>
            <address>
              <addrLine>Tour 65-66 3ème étage4, place Jussieu75005 Paris – France</addrLine>
              <country key="FR"/>
            </address>
          </desc>
          <listRelation>
            <relation active="#struct-541937" type="direct"/>
            <relation active="#struct-413221" type="indirect"/>
            <relation name="UMR7222" active="#struct-441569" type="indirect"/>
          </listRelation>
        </org>
        <org type="laboratory" xml:id="struct-541937" status="VALID">
          <idno type="IdRef">241179122</idno>
          <idno type="RNSR">200918463J</idno>
          <idno type="ROR">05neq8668</idno>
          <orgName>Institut des Systèmes Intelligents et de Robotique</orgName>
          <orgName type="acronym">ISIR</orgName>
          <date type="start">2018-01-01</date>
          <desc>
            <address>
              <addrLine>Sorbonne-Université, Boite courrier 173 4 Place JUSSIEU 75252 Paris cedex 05</addrLine>
              <country key="FR"/>
            </address>
            <ref type="url">http://www.isir.upmc.fr</ref>
          </desc>
          <listRelation>
            <relation active="#struct-413221" type="direct"/>
            <relation name="UMR7222" active="#struct-441569" type="direct"/>
          </listRelation>
        </org>
        <org type="regroupinstitution" xml:id="struct-413221" status="VALID">
          <idno type="IdRef">221333754</idno>
          <idno type="ROR">https://ror.org/02en5vm52</idno>
          <orgName>Sorbonne Université</orgName>
          <orgName type="acronym">SU</orgName>
          <date type="start">2018-01-01</date>
          <desc>
            <address>
              <addrLine>21 rue de l’École de médecine - 75006 Paris</addrLine>
              <country key="FR"/>
            </address>
            <ref type="url">http://www.sorbonne-universite.fr/</ref>
          </desc>
        </org>
        <org type="regroupinstitution" xml:id="struct-441569" status="VALID">
          <idno type="IdRef">02636817X</idno>
          <idno type="ISNI">0000000122597504</idno>
          <idno type="ROR">https://ror.org/02feahw73</idno>
          <orgName>Centre National de la Recherche Scientifique</orgName>
          <orgName type="acronym">CNRS</orgName>
          <date type="start">1939-10-19</date>
          <desc>
            <address>
              <country key="FR"/>
            </address>
            <ref type="url">https://www.cnrs.fr/</ref>
          </desc>
        </org>
      </listOrg>
      <listOrg type="projects">
        <org type="anrProject" xml:id="projanr-109431" status="VALID">
          <idno type="anr">ANR-23-IAS1-0006</idno>
          <orgName>TRaLaLaM</orgName>
          <desc>Traduire avec les Large Language Models</desc>
          <date type="start">2023</date>
        </org>
      </listOrg>
    </back>
  </text>
</TEI>