<?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-05288458</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="2026-05-24T03:48:02+02:00"/>
      </publicationStmt>
      <sourceDesc>
        <p part="N">HAL API Platform</p>
      </sourceDesc>
    </fileDesc>
  </teiHeader>
  <text>
    <body>
      <listBibl>
        <biblFull>
          <titleStmt>
            <title xml:lang="en">iKnow-audio: Integrating Knowledge Graphs with Audio-Language Models</title>
            <author role="aut">
              <persName>
                <forename type="first">Michel</forename>
                <surname>Olvera</surname>
              </persName>
              <email type="md5">493caedd1bca925de28b6ce5f047555b</email>
              <email type="domain">inria.fr</email>
              <idno type="idhal" notation="string">michel-olvera</idno>
              <idno type="idhal" notation="numeric">1416347</idno>
              <idno type="halauthorid" notation="string">1840499-1416347</idno>
              <affiliation ref="#struct-554452"/>
              <affiliation ref="#struct-554512"/>
            </author>
            <author role="aut">
              <persName>
                <forename type="first">Changhong</forename>
                <surname>Wang</surname>
              </persName>
              <email type="md5">1dd616cc8683c809c432e8bf438e38e3</email>
              <email type="domain">telecom-paris.fr</email>
              <idno type="idhal" notation="string">changhong-wang</idno>
              <idno type="idhal" notation="numeric">1237245</idno>
              <idno type="halauthorid" notation="string">2242394-1237245</idno>
              <affiliation ref="#struct-554452"/>
              <affiliation ref="#struct-554512"/>
            </author>
            <author role="aut">
              <persName>
                <forename type="first">Paraskevas</forename>
                <surname>Stamatiadis</surname>
              </persName>
              <email type="md5">7985262b421d8d9bdaa6b1e69b4824b4</email>
              <email type="domain">telecom-paris.fr</email>
              <idno type="idhal" notation="string">paraskevas-stamatiadis</idno>
              <idno type="idhal" notation="numeric">1416367</idno>
              <idno type="halauthorid" notation="string">3231616-1416367</idno>
              <affiliation ref="#struct-554452"/>
              <affiliation ref="#struct-554512"/>
            </author>
            <author role="aut">
              <persName>
                <forename type="first">Gaël</forename>
                <surname>Richard</surname>
              </persName>
              <email type="md5">c0a23dcedaef8bf8b03c039e9418997d</email>
              <email type="domain">telecom-paris.fr</email>
              <idno type="idhal" notation="string">gael-richard</idno>
              <idno type="idhal" notation="numeric">14146</idno>
              <idno type="halauthorid" notation="string">51091-14146</idno>
              <idno type="IDREF">https://www.idref.fr/094977208</idno>
              <idno type="ORCID">https://orcid.org/0000-0002-4960-0010</idno>
              <affiliation ref="#struct-554452"/>
              <affiliation ref="#struct-554512"/>
            </author>
            <author role="aut">
              <persName>
                <forename type="first">Slim</forename>
                <surname>Essid</surname>
              </persName>
              <email type="md5">2cd7c1ae65840b8ae58949cb228d354a</email>
              <email type="domain">telecom-paristech.fr</email>
              <idno type="idhal" notation="string">slimessid</idno>
              <idno type="idhal" notation="numeric">181234</idno>
              <idno type="halauthorid" notation="string">10980-181234</idno>
              <idno type="ORCID">https://orcid.org/0000-0002-0028-327X</idno>
              <idno type="GOOGLE SCHOLAR">https://scholar.google.com/citations?user=5dP_Pv0AAAAJ</idno>
              <idno type="IDREF">https://www.idref.fr/11025130X</idno>
              <affiliation ref="#struct-554452"/>
              <affiliation ref="#struct-554512"/>
              <affiliation ref="#struct-12319"/>
            </author>
            <editor role="depositor">
              <persName>
                <forename>Changhong</forename>
                <surname>Wang</surname>
              </persName>
              <email type="md5">1dd616cc8683c809c432e8bf438e38e3</email>
              <email type="domain">telecom-paris.fr</email>
            </editor>
            <funder ref="#projeurop-754042"/>
          </titleStmt>
          <editionStmt>
            <edition n="v1" type="current">
              <date type="whenSubmitted">2025-09-29 12:32:53</date>
              <date type="whenModified">2026-02-03 03:31:30</date>
              <date type="whenReleased">2025-10-03 14:56:46</date>
              <date type="whenProduced">2025-11-04</date>
              <date type="whenEndEmbargoed">2025-09-29</date>
              <ref type="file" target="https://hal.science/hal-05288458v1/document">
                <date notBefore="2025-09-29"/>
              </ref>
              <ref type="file" subtype="author" n="1" target="https://hal.science/hal-05288458v1/file/main.pdf" id="file-5288458-4533565">
                <date notBefore="2025-09-29"/>
              </ref>
            </edition>
            <respStmt>
              <resp>contributor</resp>
              <name key="1397274">
                <persName>
                  <forename>Changhong</forename>
                  <surname>Wang</surname>
                </persName>
                <email type="md5">1dd616cc8683c809c432e8bf438e38e3</email>
                <email type="domain">telecom-paris.fr</email>
              </name>
            </respStmt>
          </editionStmt>
          <publicationStmt>
            <distributor>CCSD</distributor>
            <idno type="halId">hal-05288458</idno>
            <idno type="halUri">https://hal.science/hal-05288458</idno>
            <idno type="halBibtex">olvera:hal-05288458</idno>
            <idno type="halRefHtml">&lt;i&gt;The 2025 Conference on Empirical Methods in Natural Language Processing&lt;/i&gt;, Nov 2025, Suzhou, China</idno>
            <idno type="halRef">The 2025 Conference on Empirical Methods in Natural Language Processing, Nov 2025, Suzhou, China</idno>
            <availability status="restricted">
              <licence target="https://creativecommons.org/licenses/by/4.0/">CC BY 4.0 - Attribution<ref corresp="#file-5288458-4533565"/></licence>
            </availability>
          </publicationStmt>
          <seriesStmt>
            <idno type="stamp" n="ENST">Ecole Nationale Supérieure des Télécommunications</idno>
            <idno type="stamp" n="TELECOM-PARISTECH" corresp="PARISTECH">Télécom Paris</idno>
            <idno type="stamp" n="OPENAIRE">OpenAIRE</idno>
            <idno type="stamp" n="LTCI" corresp="TELECOM-PARISTECH">Laboratoire Traitement et Communication de l'Information</idno>
            <idno type="stamp" n="IDS" corresp="TELECOM-PARISTECH">Département Image, Données, Signal</idno>
            <idno type="stamp" n="S2A" corresp="TELECOM-PARISTECH">Equipe Signal, Statistique et Apprentissage</idno>
            <idno type="stamp" n="IP_PARIS">Institut Polytechnique de Paris</idno>
            <idno type="stamp" n="INSTITUTS-TELECOM">composantes instituts telecom </idno>
            <idno type="stamp" n="INSTITUT-MINES-TELECOM">Institut Mines Telecom</idno>
            <idno type="stamp" n="IP-PARIS-DEPARTEMENT-MATHEMATIQUES">Département de Mathèmatiques</idno>
            <idno type="stamp" n="IP-PARIS-MATHEMATIQUES">IP Paris Département de Mathématiques</idno>
            <idno type="stamp" n="IP-PARIS-INFORMATION-COMMUNICATION-ELECTRONIQUE">IP Paris Département d'Information, de Communication et d'Électronique</idno>
            <idno type="stamp" n="IP-PARIS-INFORMATIQUE-DONNEES-ET-IA">IP Paris Département d'Informatique, de Données et d'IA</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="0">No</note>
          </notesStmt>
          <sourceDesc>
            <biblStruct>
              <analytic>
                <title xml:lang="en">iKnow-audio: Integrating Knowledge Graphs with Audio-Language Models</title>
                <author role="aut">
                  <persName>
                    <forename type="first">Michel</forename>
                    <surname>Olvera</surname>
                  </persName>
                  <email type="md5">493caedd1bca925de28b6ce5f047555b</email>
                  <email type="domain">inria.fr</email>
                  <idno type="idhal" notation="string">michel-olvera</idno>
                  <idno type="idhal" notation="numeric">1416347</idno>
                  <idno type="halauthorid" notation="string">1840499-1416347</idno>
                  <affiliation ref="#struct-554452"/>
                  <affiliation ref="#struct-554512"/>
                </author>
                <author role="aut">
                  <persName>
                    <forename type="first">Changhong</forename>
                    <surname>Wang</surname>
                  </persName>
                  <email type="md5">1dd616cc8683c809c432e8bf438e38e3</email>
                  <email type="domain">telecom-paris.fr</email>
                  <idno type="idhal" notation="string">changhong-wang</idno>
                  <idno type="idhal" notation="numeric">1237245</idno>
                  <idno type="halauthorid" notation="string">2242394-1237245</idno>
                  <affiliation ref="#struct-554452"/>
                  <affiliation ref="#struct-554512"/>
                </author>
                <author role="aut">
                  <persName>
                    <forename type="first">Paraskevas</forename>
                    <surname>Stamatiadis</surname>
                  </persName>
                  <email type="md5">7985262b421d8d9bdaa6b1e69b4824b4</email>
                  <email type="domain">telecom-paris.fr</email>
                  <idno type="idhal" notation="string">paraskevas-stamatiadis</idno>
                  <idno type="idhal" notation="numeric">1416367</idno>
                  <idno type="halauthorid" notation="string">3231616-1416367</idno>
                  <affiliation ref="#struct-554452"/>
                  <affiliation ref="#struct-554512"/>
                </author>
                <author role="aut">
                  <persName>
                    <forename type="first">Gaël</forename>
                    <surname>Richard</surname>
                  </persName>
                  <email type="md5">c0a23dcedaef8bf8b03c039e9418997d</email>
                  <email type="domain">telecom-paris.fr</email>
                  <idno type="idhal" notation="string">gael-richard</idno>
                  <idno type="idhal" notation="numeric">14146</idno>
                  <idno type="halauthorid" notation="string">51091-14146</idno>
                  <idno type="IDREF">https://www.idref.fr/094977208</idno>
                  <idno type="ORCID">https://orcid.org/0000-0002-4960-0010</idno>
                  <affiliation ref="#struct-554452"/>
                  <affiliation ref="#struct-554512"/>
                </author>
                <author role="aut">
                  <persName>
                    <forename type="first">Slim</forename>
                    <surname>Essid</surname>
                  </persName>
                  <email type="md5">2cd7c1ae65840b8ae58949cb228d354a</email>
                  <email type="domain">telecom-paristech.fr</email>
                  <idno type="idhal" notation="string">slimessid</idno>
                  <idno type="idhal" notation="numeric">181234</idno>
                  <idno type="halauthorid" notation="string">10980-181234</idno>
                  <idno type="ORCID">https://orcid.org/0000-0002-0028-327X</idno>
                  <idno type="GOOGLE SCHOLAR">https://scholar.google.com/citations?user=5dP_Pv0AAAAJ</idno>
                  <idno type="IDREF">https://www.idref.fr/11025130X</idno>
                  <affiliation ref="#struct-554452"/>
                  <affiliation ref="#struct-554512"/>
                  <affiliation ref="#struct-12319"/>
                </author>
              </analytic>
              <monogr>
                <meeting>
                  <title>The 2025 Conference on Empirical Methods in Natural Language Processing</title>
                  <date type="start">2025-11-04</date>
                  <date type="end">2025-11-09</date>
                  <settlement>Suzhou</settlement>
                  <country key="CN">China</country>
                </meeting>
                <imprint/>
              </monogr>
            </biblStruct>
          </sourceDesc>
          <profileDesc>
            <langUsage>
              <language ident="en">English</language>
            </langUsage>
            <textClass>
              <keywords scheme="author">
                <term xml:lang="en">Knowledge graph</term>
                <term xml:lang="en">Zero-shot audio classification</term>
                <term xml:lang="en">Audio-language models</term>
              </keywords>
              <classCode scheme="halDomain" n="info.info-sd">Computer Science [cs]/Sound [cs.SD]</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>Contrastive Language–Audio Pretraining (CLAP) models learn by aligning audio and text in a shared embedding space, enabling powerful zero-shot recognition. However, their performance is highly sensitive to prompt formulation and language nuances, and they often inherit semantic ambiguities and spurious correlations from noisy pretraining data. While prior work has explored prompt engineering, adapters, and prefix tuning to address these limitations, the use of structured prior knowledge remains largely unexplored. We present iKnow-audio, a framework that integrates knowledge graphs with audio-language models to provide robust semantic grounding. iKnow-audio builds on the Audio-centric Knowledge Graph (AKG), which encodes ontological relations comprising semantic, causal, and taxonomic connections reflective of everyday sound scenes and events. By training knowlege graph embedding models on the AKG and refining CLAP predictions through this structured knowledge, iKnow-audio improves disambiguation of acoustically similar sounds and reduces reliance on prompt engineering. Comprehensive zero-shot evaluations across six benchmark datasets demonstrate consistent gains over baseline CLAP, supported by embedding-space analyses that highlight improved relational grounding. Resources are publicly available at https://github.com/michelolzam/iknow-audio.</p>
            </abstract>
          </profileDesc>
        </biblFull>
      </listBibl>
    </body>
    <back>
      <listOrg type="structures">
        <org type="researchteam" xml:id="struct-554452" status="VALID">
          <orgName>Signal, Statistique et Apprentissage</orgName>
          <orgName type="acronym">S2A</orgName>
          <date type="start">2019-02-22</date>
          <desc>
            <address>
              <addrLine>Télécom Paris 19 Place Marguerite Perey 91120 Palaiseau</addrLine>
              <country key="FR"/>
            </address>
            <ref type="url">https://www.telecom-paris.fr/fr/recherche/laboratoires/laboratoire-traitement-et-communication-de-linformation-ltci/les-equipes-de-recherche/signal-statistique-et-apprentissage-s2a</ref>
          </desc>
          <listRelation>
            <relation active="#struct-484335" type="direct"/>
            <relation active="#struct-302102" type="indirect"/>
            <relation active="#struct-1048346" type="indirect"/>
            <relation active="#struct-563936" type="indirect"/>
          </listRelation>
        </org>
        <org type="department" xml:id="struct-554512" status="VALID">
          <orgName>Département Images, Données, Signal</orgName>
          <orgName type="acronym">IDS</orgName>
          <date type="start">2017-01-01</date>
          <desc>
            <address>
              <addrLine>46, rue Barrault 75013 Paris ; 15 Place Marguerite Perey 91120 Palaiseau (depuis oct 2019)</addrLine>
              <country key="FR"/>
            </address>
            <ref type="url">https://www.telecom-paris.fr/fr/recherche/laboratoires/laboratoire-traitement-et-communication-de-linformation-ltci/les-departements-denseignement-et-recherche</ref>
          </desc>
          <listRelation>
            <relation active="#struct-300362" type="direct"/>
          </listRelation>
        </org>
        <org type="institution" xml:id="struct-12319" status="VALID">
          <orgName>NVIDIA</orgName>
          <orgName type="acronym">NVIDIA</orgName>
          <desc>
            <address>
              <country key="US"/>
            </address>
            <ref type="url">http://www.nvidia.com/content/global/global.php</ref>
          </desc>
        </org>
        <org type="laboratory" xml:id="struct-484335" status="VALID">
          <idno type="IdRef">162384270</idno>
          <idno type="ISNI">0000 0000 9194 9502</idno>
          <idno type="RNSR">200319327Z</idno>
          <idno type="ROR">https://ror.org/057er4c39</idno>
          <orgName>Laboratoire Traitement et Communication de l'Information</orgName>
          <orgName type="acronym">LTCI</orgName>
          <date type="start">2017-01-01</date>
          <desc>
            <address>
              <addrLine>Télécom Paris 19 Place Marguerite Perey 91120 PALAISEAU</addrLine>
              <country key="FR"/>
            </address>
            <ref type="url">https://www.telecom-paris.fr/fr/recherche/laboratoires/laboratoire-traitement-et-communication-de-linformation-ltci</ref>
          </desc>
          <listRelation>
            <relation active="#struct-302102" type="direct"/>
            <relation active="#struct-1048346" type="direct"/>
            <relation active="#struct-563936" type="indirect"/>
          </listRelation>
        </org>
        <org type="regroupinstitution" xml:id="struct-302102" status="VALID">
          <idno type="IdRef">192427156</idno>
          <idno type="ISNI">000000012202567X</idno>
          <idno type="ROR">https://ror.org/025vp2923</idno>
          <idno type="Wikidata">Q27962533</idno>
          <orgName>Institut Mines-Télécom [Paris]</orgName>
          <orgName type="acronym">IMT</orgName>
          <date type="start">2012-03-01</date>
          <desc>
            <address>
              <addrLine>19 Place Marguerite Perey, 91120 Palaiseau</addrLine>
              <country key="FR"/>
            </address>
            <ref type="url">https://www.imt.fr/</ref>
          </desc>
        </org>
        <org type="institution" xml:id="struct-1048346" status="VALID">
          <idno type="IdRef">026375273</idno>
          <idno type="ISNI">0000 0001 2108 2779</idno>
          <idno type="ROR">https://ror.org/01naq7912</idno>
          <orgName>Télécom Paris</orgName>
          <date type="start">2019-06-12</date>
          <desc>
            <address>
              <addrLine>19 Place Marguerite Perey 91120 Palaiseau</addrLine>
              <country key="FR"/>
            </address>
            <ref type="url">https://www.telecom-paris.fr</ref>
          </desc>
          <listRelation>
            <relation active="#struct-302102" type="direct"/>
            <relation active="#struct-563936" type="direct"/>
          </listRelation>
        </org>
        <org type="regroupinstitution" xml:id="struct-563936" status="VALID">
          <idno type="IdRef">238327159</idno>
          <idno type="ISNI">0000000502717600</idno>
          <idno type="ROR">https://ror.org/042tfbd02</idno>
          <idno type="Wikidata">Q48759778</idno>
          <orgName>Institut Polytechnique de Paris</orgName>
          <orgName type="acronym">IP Paris</orgName>
          <date type="start">2019-06-02</date>
          <desc>
            <address>
              <addrLine>Route de Saclay, 91120 Palaiseau Cedex, France</addrLine>
              <country key="FR"/>
            </address>
            <ref type="url">https://www.ip-paris.fr</ref>
          </desc>
        </org>
        <org type="institution" xml:id="struct-300362" status="VALID">
          <idno type="ROR">https://ror.org/01naq7912</idno>
          <orgName>Télécom ParisTech</orgName>
          <date type="start">2008-01-01</date>
          <date type="end">2019-06-11</date>
          <desc>
            <address>
              <addrLine>46 rue Barrault 75634 Paris Cedex 13</addrLine>
              <country key="FR"/>
            </address>
            <ref type="url">https://www.telecom-paris.fr</ref>
          </desc>
        </org>
      </listOrg>
      <listOrg type="projects">
        <org type="europeanProject" xml:id="projeurop-754042" status="VALID">
          <idno type="number">101052978</idno>
          <idno type="program">ERC-2021-ADG</idno>
          <idno type="call">ERC-2021-ADG</idno>
          <orgName>HI-Audio</orgName>
          <desc>Hybrid and Interpretable Deep neural audio machines</desc>
          <date type="start">2022-10-01</date>
          <date type="end">2027-09-30</date>
        </org>
      </listOrg>
    </back>
  </text>
</TEI>