{"id":"https://openalex.org/W4416033506","doi":"https://doi.org/10.18653/v1/2025.wmt-1.72","title":"UvA-MT at WMT25 Evaluation Task: LLM Uncertainty as a Proxy for Translation Quality","display_name":"UvA-MT at WMT25 Evaluation Task: LLM Uncertainty as a Proxy for Translation Quality","publication_year":2025,"publication_date":"2025-01-01","ids":{"openalex":"https://openalex.org/W4416033506","doi":"https://doi.org/10.18653/v1/2025.wmt-1.72"},"language":"en","primary_location":{"id":"doi:10.18653/v1/2025.wmt-1.72","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.wmt-1.72","pdf_url":"https://aclanthology.org/2025.wmt-1.72.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-1.72.pdf","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5101667547","display_name":"Di Wu","orcid":"https://orcid.org/0000-0003-2092-8650"},"institutions":[{"id":"https://openalex.org/I887064364","display_name":"University of Amsterdam","ror":"https://ror.org/04dkp9463","country_code":"NL","type":"education","lineage":["https://openalex.org/I887064364"]}],"countries":["NL"],"is_corresponding":false,"raw_author_name":"Di Wu","raw_affiliation_strings":["Language Technology Lab University of Amsterdam"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Language Technology Lab University of Amsterdam","institution_ids":["https://openalex.org/I887064364"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5109059955","display_name":"Christof Monz","orcid":null},"institutions":[{"id":"https://openalex.org/I887064364","display_name":"University of Amsterdam","ror":"https://ror.org/04dkp9463","country_code":"NL","type":"education","lineage":["https://openalex.org/I887064364"]}],"countries":["NL"],"is_corresponding":false,"raw_author_name":"Christof Monz","raw_affiliation_strings":["Language Technology Lab University of Amsterdam"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Language Technology Lab University of Amsterdam","institution_ids":["https://openalex.org/I887064364"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I887064364"],"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":"974","last_page":"983"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10181","display_name":"Natural Language Processing Techniques","score":0.9327999949455261,"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":0.9327999949455261,"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/T13629","display_name":"Text Readability and Simplification","score":0.009600000455975533,"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":0.006899999920278788,"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"}}],"keywords":[{"id":"https://openalex.org/keywords/proxy","display_name":"Proxy (statistics)","score":0.511900007724762},{"id":"https://openalex.org/keywords/quality","display_name":"Quality (philosophy)","score":0.4796000123023987},{"id":"https://openalex.org/keywords/quality-assessment","display_name":"Quality assessment","score":0.35359999537467957},{"id":"https://openalex.org/keywords/quality-assurance","display_name":"Quality assurance","score":0.2994999885559082},{"id":"https://openalex.org/keywords/data-quality","display_name":"Data quality","score":0.295199990272522}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5537999868392944},{"id":"https://openalex.org/C2780148112","wikidata":"https://www.wikidata.org/wiki/Q1432581","display_name":"Proxy (statistics)","level":2,"score":0.511900007724762},{"id":"https://openalex.org/C2779530757","wikidata":"https://www.wikidata.org/wiki/Q1207505","display_name":"Quality (philosophy)","level":2,"score":0.4796000123023987},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.35899999737739563},{"id":"https://openalex.org/C3020001037","wikidata":"https://www.wikidata.org/wiki/Q836575","display_name":"Quality assessment","level":3,"score":0.35359999537467957},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.34850001335144043},{"id":"https://openalex.org/C106436119","wikidata":"https://www.wikidata.org/wiki/Q836575","display_name":"Quality assurance","level":3,"score":0.2994999885559082},{"id":"https://openalex.org/C24756922","wikidata":"https://www.wikidata.org/wiki/Q1757694","display_name":"Data quality","level":3,"score":0.295199990272522},{"id":"https://openalex.org/C149364088","wikidata":"https://www.wikidata.org/wiki/Q185917","display_name":"Translation (biology)","level":4,"score":0.2883000075817108},{"id":"https://openalex.org/C112930515","wikidata":"https://www.wikidata.org/wiki/Q4389547","display_name":"Risk analysis (engineering)","level":1,"score":0.2460000067949295}],"mesh":[],"locations_count":3,"locations":[{"id":"doi:10.18653/v1/2025.wmt-1.72","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.wmt-1.72","pdf_url":"https://aclanthology.org/2025.wmt-1.72.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:openaire_cris_publications/0cdcb0c7-b1e2-488c-81aa-b1f5ff385c4d","is_oa":true,"landing_page_url":"https://handle.uba.uva.nl/personal/pure/en/publications/uvamt-at-wmt25-evaluation-task(0cdcb0c7-b1e2-488c-81aa-b1f5ff385c4d).html","pdf_url":"https://pure.uva.nl/ws/files/306463005/2025.wmt-1.72.pdf","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":"Wu, D & Monz, C 2025, UvA-MT at WMT25 Evaluation Task : LLM Uncertainty as a Proxy for Translation Quality. 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. 974-983, 10th Conference on Machine Translation, WMT 2025, Suzhou, China, 8/11/25. https://doi.org/10.18653/v1/2025.wmt-1.72","raw_type":"info:eu-repo/semantics/conferenceObject"},{"id":"pmh:oai:dare.uva.nl:publications/0cdcb0c7-b1e2-488c-81aa-b1f5ff385c4d","is_oa":true,"landing_page_url":"https://hdl.handle.net/11245.1/0cdcb0c7-b1e2-488c-81aa-b1f5ff385c4d","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":"Wu, D & Monz, C 2025, UvA-MT at WMT25 Evaluation Task : LLM Uncertainty as a Proxy for Translation Quality. 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. 974-983, 10th Conference on Machine Translation, WMT 2025, Suzhou, China, 8/11/25. https://doi.org/10.18653/v1/2025.wmt-1.72","raw_type":"info:eu-repo/semantics/conferenceObject"}],"best_oa_location":{"id":"doi:10.18653/v1/2025.wmt-1.72","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.wmt-1.72","pdf_url":"https://aclanthology.org/2025.wmt-1.72.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":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4416033506.pdf","grobid_xml":"https://content.openalex.org/works/W4416033506.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"This":[0],"year,":[1],"we":[2,57,77,165],"focus":[3],"exclusively":[4],"on":[5],"using":[6],"the":[7,34,37,45,79,87,156,181],"uncertainty":[8,101,141],"quantification":[9,142],"as":[10,21,136,149],"a":[11,22,150],"proxy":[12],"for":[13,155],"translation":[14,68,89],"quality.While":[15],"this":[16,120],"has":[17],"traditionally":[18],"been":[19,31],"regarded":[20],"form":[23],"of":[24,36,47],"unsupervised":[25,54],"quality":[26,74,112],"estimation,":[27],"such":[28,135],"signals":[29,102],"have":[30],"overlooked":[32],"in":[33,44,50,85],"design":[35],"current":[38],"metric":[39,127,157],"models-we":[40],"show":[41,97],"their":[42],"value":[43],"context":[46],"LLMs.More":[48],"specifically,":[49],"contrast":[51],"to":[52,66,70,83,172,177],"conventional":[53],"QE":[55,91,121],"methods,":[56],"apply":[58],"recent":[59],"calibration":[60,116],"technology":[61,117],"(Wu":[62],"et":[63],"al.,":[64],"2025b)":[65],"adjust":[67],"likelihoods":[69],"better":[71],"align":[72],"with":[73,132],"signals,":[75],"and":[76,90,114,152],"use":[78],"single":[80],"resulting":[81],"model":[82,182],"participate":[84],"both":[86],"general":[88],"tracks":[92],"at":[93],"WMT25.Our":[94],"offline":[95],"experiments":[96],"some":[98],"advantages:":[99],"1)":[100],"extracted":[103],"from":[104,145],"LLMs,":[105,146],"like":[106],"Tower":[107],"or":[108],"Gemma-3,":[109],"provide":[110],"accurate":[111],"predictions;":[113],"2)":[115],"further":[118],"improves":[119],"performance,":[122],"sometimes":[123],"even":[124],"surpassing":[125],"certain":[126],"models":[128],"that":[129,140],"were":[130],"trained":[131],"human":[133],"annotations,":[134],"CometKiwi.We":[137],"therefore":[138],"argue":[139],"(confidence),":[143],"especially":[144],"can":[147],"serve":[148],"strong":[151],"complementary":[153],"signal":[154],"design,":[158],"particularly":[159],"when":[160],"human-annotated":[161],"data":[162],"are":[163],"lacking.However,":[164],"also":[166],"identify":[167],"limitations,":[168],"i.e.,":[169],"its":[170],"tendency":[171],"assign":[173],"disproportionately":[174],"higher":[175],"scores":[176],"hypotheses":[178],"generated":[179],"by":[180],"itself.":[183]},"counts_by_year":[],"updated_date":"2026-07-14T23:27:15.235271","created_date":"2025-11-08T00:00:00"}