{"id":"https://openalex.org/W4416033721","doi":"https://doi.org/10.18653/v1/2025.wmt-1.1","title":"An Empirical Analysis of Machine Translation for Expanding Multilingual Benchmarks","display_name":"An Empirical Analysis of Machine Translation for Expanding Multilingual Benchmarks","publication_year":2025,"publication_date":"2025-01-01","ids":{"openalex":"https://openalex.org/W4416033721","doi":"https://doi.org/10.18653/v1/2025.wmt-1.1"},"language":"en","primary_location":{"id":"doi:10.18653/v1/2025.wmt-1.1","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.wmt-1.1","pdf_url":"https://aclanthology.org/2025.wmt-1.1.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the Tenth Conference on Machine Translation","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2025.wmt-1.1.pdf","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5037424359","display_name":"Sara Rajaee","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Sara Rajaee","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5109855717","display_name":"Rochelle Choenni","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Rochelle Choenni","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5016184654","display_name":"Ekaterina Shutova","orcid":"https://orcid.org/0009-0003-6664-4474"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ekaterina Shutova","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5109059955","display_name":"Christof Monz","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Christof Monz","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"1","last_page":"30"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10181","display_name":"Natural Language Processing Techniques","score":0.8427000045776367,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},"topics":[{"id":"https://openalex.org/T10181","display_name":"Natural Language Processing Techniques","score":0.8427000045776367,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10759","display_name":"Translation Studies and Practices","score":0.007699999958276749,"subfield":{"id":"https://openalex.org/subfields/1203","display_name":"Language and Linguistics"},"field":{"id":"https://openalex.org/fields/12","display_name":"Arts and Humanities"},"domain":{"id":"https://openalex.org/domains/2","display_name":"Social Sciences"}},{"id":"https://openalex.org/T13629","display_name":"Text Readability and Simplification","score":0.004100000020116568,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/machine-translation","display_name":"Machine translation","score":0.5356000065803528},{"id":"https://openalex.org/keywords/translation","display_name":"Translation (biology)","score":0.3296000063419342},{"id":"https://openalex.org/keywords/feature","display_name":"Feature (linguistics)","score":0.29249998927116394},{"id":"https://openalex.org/keywords/interpretation","display_name":"Interpretation (philosophy)","score":0.27970001101493835},{"id":"https://openalex.org/keywords/empirical-research","display_name":"Empirical research","score":0.2718000113964081}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7113999724388123},{"id":"https://openalex.org/C203005215","wikidata":"https://www.wikidata.org/wiki/Q79798","display_name":"Machine translation","level":2,"score":0.5356000065803528},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.5250999927520752},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4788999855518341},{"id":"https://openalex.org/C149364088","wikidata":"https://www.wikidata.org/wiki/Q185917","display_name":"Translation (biology)","level":4,"score":0.3296000063419342},{"id":"https://openalex.org/C2776401178","wikidata":"https://www.wikidata.org/wiki/Q12050496","display_name":"Feature (linguistics)","level":2,"score":0.29249998927116394},{"id":"https://openalex.org/C527412718","wikidata":"https://www.wikidata.org/wiki/Q855395","display_name":"Interpretation (philosophy)","level":2,"score":0.27970001101493835},{"id":"https://openalex.org/C120936955","wikidata":"https://www.wikidata.org/wiki/Q2155640","display_name":"Empirical research","level":2,"score":0.2718000113964081},{"id":"https://openalex.org/C41895202","wikidata":"https://www.wikidata.org/wiki/Q8162","display_name":"Linguistics","level":1,"score":0.2694999873638153},{"id":"https://openalex.org/C184337299","wikidata":"https://www.wikidata.org/wiki/Q1437428","display_name":"Semantics (computer science)","level":2,"score":0.25839999318122864},{"id":"https://openalex.org/C155092808","wikidata":"https://www.wikidata.org/wiki/Q182557","display_name":"Computational linguistics","level":2,"score":0.25450000166893005},{"id":"https://openalex.org/C199360897","wikidata":"https://www.wikidata.org/wiki/Q9143","display_name":"Programming language","level":1,"score":0.25450000166893005}],"mesh":[],"locations_count":3,"locations":[{"id":"doi:10.18653/v1/2025.wmt-1.1","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.wmt-1.1","pdf_url":"https://aclanthology.org/2025.wmt-1.1.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the Tenth Conference on Machine Translation","raw_type":"proceedings-article"},{"id":"pmh:oai:dare.uva.nl:publications/8e00000d-fa82-4f16-b7f7-93cfd25914a9","is_oa":true,"landing_page_url":"https://handle.uba.uva.nl/personal/pure/en/publications/an-empirical-analysis-of-machine-translation-for-expanding-multilingual-benchmarks(8e00000d-fa82-4f16-b7f7-93cfd25914a9).html","pdf_url":null,"source":{"id":"https://openalex.org/S4306400088","display_name":"UvA-DARE (University of Amsterdam)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I887064364","host_organization_name":"University of Amsterdam","host_organization_lineage":["https://openalex.org/I887064364"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Rajaee, S, Choenni, R, Shutova, E & Monz, C 2025, An Empirical Analysis of Machine Translation for Expanding Multilingual Benchmarks. in B Haddow, T Kocmi, P Koehn & C Monz (eds), Tenth Conference on Machine Translation : Proceedings of the Conference : WMT 2025 : November 8-9, 2025. Kerrville, TX, pp. 1-30, 10th Conference on Machine Translation, WMT 2025, Suzhou, China, 8/11/25. https://doi.org/10.18653/v1/2025.wmt-1.1","raw_type":"info:eu-repo/semantics/conferenceObject"},{"id":"pmh:oai:dare.uva.nl:openaire_cris_publications/8e00000d-fa82-4f16-b7f7-93cfd25914a9","is_oa":true,"landing_page_url":"https://hdl.handle.net/11245.1/8e00000d-fa82-4f16-b7f7-93cfd25914a9","pdf_url":null,"source":{"id":"https://openalex.org/S4306400088","display_name":"UvA-DARE (University of Amsterdam)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I887064364","host_organization_name":"University of Amsterdam","host_organization_lineage":["https://openalex.org/I887064364"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"Rajaee, S, Choenni, R, Shutova, E & Monz, C 2025, An Empirical Analysis of Machine Translation for Expanding Multilingual Benchmarks. in B Haddow, T Kocmi, P Koehn & C Monz (eds), Tenth Conference on Machine Translation : Proceedings of the Conference : WMT 2025 : November 8-9, 2025. Kerrville, TX, pp. 1-30, 10th Conference on Machine Translation, WMT 2025, Suzhou, China, 8/11/25. https://doi.org/10.18653/v1/2025.wmt-1.1","raw_type":"info:eu-repo/semantics/conferenceObject"}],"best_oa_location":{"id":"doi:10.18653/v1/2025.wmt-1.1","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.wmt-1.1","pdf_url":"https://aclanthology.org/2025.wmt-1.1.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the Tenth Conference on Machine Translation","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4416033721.pdf"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"The":[0,19],"rapid":[1],"advancement":[2],"of":[3,33,49,91],"large":[4],"language":[5,121],"models":[6],"(LLMs)":[7],"has":[8],"introduced":[9],"new":[10],"challenges":[11],"in":[12,26,59,123,162],"their":[13],"evaluation,":[14],"particularly":[15],"for":[16,57,147],"multilingual":[17,93,124,149],"settings.":[18],"limited":[20],"evaluation":[21,58,94,106],"data":[22],"are":[23,65],"more":[24],"pronounced":[25],"low-resource":[27,132],"languages":[28,80],"due":[29],"to":[30,85,127,159],"the":[31,47,87],"scarcity":[32],"professional":[34],"annotators,":[35],"hindering":[36],"fair":[37],"progress":[38,161],"across":[39],"languages.":[40,133,164],"In":[41],"this":[42],"work,":[43],"we":[44,72],"systematically":[45],"investigate":[46],"viability":[48],"using":[50],"machine":[51,138],"translation":[52,70,139],"(MT)":[53],"as":[54],"a":[55,68,143,156],"proxy":[56],"scenarios":[60],"where":[61],"human-annotated":[62],"test":[63],"sets":[64],"unavailable.":[66],"Leveraging":[67],"state-of-the-art":[69],"model,":[71],"translate":[73],"datasets":[74,125],"from":[75],"four":[76],"tasks":[77],"into":[78],"198":[79],"and":[81,89,109],"employ":[82],"these":[83],"translations":[84],"assess":[86],"quality":[88],"robustness":[90],"MT-based":[92,105],"under":[95],"different":[96],"setups.":[97],"We":[98,134],"analyze":[99],"task-specific":[100],"error":[101],"patterns,":[102],"identifying":[103],"when":[104,110],"is":[107,140],"reliable":[108,145],"it":[111],"produces":[112],"misleading":[113],"results.":[114],"Our":[115],"translated":[116],"benchmark":[117],"reveals":[118],"that":[119,136],"current":[120],"selections":[122],"tend":[126],"overestimate":[128],"LLM":[129],"performance":[130],"on":[131],"conclude":[135],"although":[137],"not":[141],"yet":[142],"fully":[144],"method":[146],"evaluating":[148],"models,":[150],"overlooking":[151],"its":[152],"potential":[153],"means":[154],"missing":[155],"valuable":[157],"opportunity":[158],"track":[160],"non-English":[163]},"counts_by_year":[],"updated_date":"2026-07-14T23:27:15.235271","created_date":"2025-11-08T00:00:00"}