{"id":"https://openalex.org/W4412888936","doi":"https://doi.org/10.18653/v1/2025.findings-acl.31","title":"CRPO: Confidence-Reward Driven Preference Optimization for Machine Translation","display_name":"CRPO: Confidence-Reward Driven Preference Optimization for Machine Translation","publication_year":2025,"publication_date":"2025-01-01","ids":{"openalex":"https://openalex.org/W4412888936","doi":"https://doi.org/10.18653/v1/2025.findings-acl.31"},"language":"en","primary_location":{"id":"doi:10.18653/v1/2025.findings-acl.31","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-acl.31","pdf_url":"https://aclanthology.org/2025.findings-acl.31.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":"Findings of the Association for Computational Linguistics: ACL 2025","raw_type":"proceedings-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2025.findings-acl.31.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5090108980","display_name":"Guofeng Cui","orcid":"https://orcid.org/0000-0001-7152-8185"},"institutions":[],"countries":[],"is_corresponding":true,"raw_author_name":"Guofeng Cui","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5042680345","display_name":"Pichao Wang","orcid":"https://orcid.org/0000-0002-1430-0237"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Pichao Wang","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100355692","display_name":"Yang Liu","orcid":"https://orcid.org/0000-0001-7300-9215"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yang Liu","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5030124388","display_name":"Zemian Ke","orcid":"https://orcid.org/0000-0002-8616-7498"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zemian Ke","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100511786","display_name":"Zhu Liu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhu Liu","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5050133239","display_name":"Vimal Bhat","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Vimal Bhat","raw_affiliation_strings":[],"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":6,"corresponding_author_ids":["https://openalex.org/A5090108980"],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":{"value":0.09967991,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"560","last_page":"574"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10181","display_name":"Natural Language Processing Techniques","score":0.9876999855041504,"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":0.9876999855041504,"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/T10215","display_name":"Semantic Web and Ontologies","score":0.9714000225067139,"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/preference","display_name":"Preference","score":0.735227108001709},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.6993438005447388},{"id":"https://openalex.org/keywords/machine-translation","display_name":"Machine translation","score":0.679732620716095},{"id":"https://openalex.org/keywords/translation","display_name":"Translation (biology)","score":0.534630537033081},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.5094013810157776},{"id":"https://openalex.org/keywords/machine-learning","display_name":"Machine learning","score":0.3277657926082611},{"id":"https://openalex.org/keywords/statistics","display_name":"Statistics","score":0.19679152965545654},{"id":"https://openalex.org/keywords/mathematics","display_name":"Mathematics","score":0.13738569617271423}],"concepts":[{"id":"https://openalex.org/C2781249084","wikidata":"https://www.wikidata.org/wiki/Q908656","display_name":"Preference","level":2,"score":0.735227108001709},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6993438005447388},{"id":"https://openalex.org/C203005215","wikidata":"https://www.wikidata.org/wiki/Q79798","display_name":"Machine translation","level":2,"score":0.679732620716095},{"id":"https://openalex.org/C149364088","wikidata":"https://www.wikidata.org/wiki/Q185917","display_name":"Translation (biology)","level":4,"score":0.534630537033081},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5094013810157776},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.3277657926082611},{"id":"https://openalex.org/C105795698","wikidata":"https://www.wikidata.org/wiki/Q12483","display_name":"Statistics","level":1,"score":0.19679152965545654},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.13738569617271423},{"id":"https://openalex.org/C55493867","wikidata":"https://www.wikidata.org/wiki/Q7094","display_name":"Biochemistry","level":1,"score":0.0},{"id":"https://openalex.org/C105580179","wikidata":"https://www.wikidata.org/wiki/Q188928","display_name":"Messenger RNA","level":3,"score":0.0},{"id":"https://openalex.org/C104317684","wikidata":"https://www.wikidata.org/wiki/Q7187","display_name":"Gene","level":2,"score":0.0},{"id":"https://openalex.org/C185592680","wikidata":"https://www.wikidata.org/wiki/Q2329","display_name":"Chemistry","level":0,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2025.findings-acl.31","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-acl.31","pdf_url":"https://aclanthology.org/2025.findings-acl.31.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":"Findings of the Association for Computational Linguistics: ACL 2025","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2025.findings-acl.31","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-acl.31","pdf_url":"https://aclanthology.org/2025.findings-acl.31.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":"Findings of the Association for Computational Linguistics: ACL 2025","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4412888936.pdf","grobid_xml":"https://content.openalex.org/works/W4412888936.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W2961085424","https://openalex.org/W4306674287","https://openalex.org/W4387369504","https://openalex.org/W4394896187","https://openalex.org/W3170094116","https://openalex.org/W4386462264","https://openalex.org/W2883671469","https://openalex.org/W3107602296","https://openalex.org/W2728761353","https://openalex.org/W4364306694"],"abstract_inverted_index":{"Large":[0],"language":[1,10],"models":[2,111],"(LLMs)":[3],"have":[4],"shown":[5],"great":[6],"potential":[7],"in":[8,131],"natural":[9],"processing":[11],"tasks,":[12],"but":[13,50],"their":[14],"application":[15],"to":[16,23,80,98,109],"machine":[17],"translation":[18,133],"(MT)":[19],"remains":[20],"challenging":[21,87],"due":[22],"pretraining":[24],"on":[25,55],"English-centric":[26],"data":[27,82,136],"and":[28,46,128,135],"the":[29,56,91],"complexity":[30],"of":[31,58],"reinforcement":[32],"learning":[33],"from":[34],"human":[35],"feedback":[36],"(RLHF).Direct":[37],"Preference":[38,67],"Optimization":[39,68],"(DPO)":[40],"has":[41],"emerged":[42],"as":[43,125],"a":[44,70],"simpler":[45],"more":[47,99],"efficient":[48],"alternative,":[49],"its":[51,115],"performance":[52],"depends":[53],"heavily":[54],"quality":[57],"preference":[59],"data.To":[60],"address":[61],"this,":[62],"we":[63],"propose":[64],"Confidence-Reward":[65],"driven":[66],"(CRPO),":[69],"novel":[71],"method":[72],"that":[73,119],"combines":[74],"reward":[75],"scores":[76],"with":[77],"model":[78,92],"confidence":[79],"improve":[81],"selection":[83],"for":[84,104],"fine-tuning.CRPO":[85],"selects":[86],"sentence":[88],"pairs":[89],"where":[90],"is":[93],"uncertain":[94],"or":[95],"underperforms,":[96],"leading":[97],"effective":[100],"learning.While":[101],"primarily":[102],"designed":[103],"LLMs,":[105],"CRPO":[106,120],"also":[107],"generalizes":[108],"encoder-decoder":[110],"like":[112],"NLLB,":[113],"demonstrating":[114],"versatility.Empirical":[116],"results":[117],"show":[118],"outperforms":[121],"existing":[122],"methods":[123],"such":[124],"RS-DPO,":[126],"RSO":[127],"MBR":[129],"score":[130],"both":[132],"accuracy":[134],"efficiency.":[137]},"counts_by_year":[],"updated_date":"2026-03-11T06:11:40.159057","created_date":"2025-10-10T00:00:00"}