{"id":"https://openalex.org/W4412888217","doi":"https://doi.org/10.18653/v1/2025.findings-acl.700","title":"ReasonerRank: Redefining Language Model Evaluation with Ground-Truth-Free Ranking Frameworks","display_name":"ReasonerRank: Redefining Language Model Evaluation with Ground-Truth-Free Ranking Frameworks","publication_year":2025,"publication_date":"2025-01-01","ids":{"openalex":"https://openalex.org/W4412888217","doi":"https://doi.org/10.18653/v1/2025.findings-acl.700"},"language":"en","primary_location":{"id":"doi:10.18653/v1/2025.findings-acl.700","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-acl.700","pdf_url":"https://aclanthology.org/2025.findings-acl.700.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.700.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":null,"display_name":"Jiamu Zhang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jiamu Zhang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5067607368","display_name":"Jiayi Yuan","orcid":"https://orcid.org/0009-0007-0121-4990"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jiayi Yuan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5057790510","display_name":"Andrew Wen","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Andrew Wen","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5015151995","display_name":"Huong Thi Le","orcid":"https://orcid.org/0000-0002-2913-7065"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Hoang Anh Duy Le","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5048791081","display_name":"Yu-Neng Chuang","orcid":"https://orcid.org/0000-0002-7492-0817"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yu-Neng Chuang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5027525032","display_name":"Soo-Hyun Choi","orcid":"https://orcid.org/0000-0001-5768-9978"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Soo-Hyun Choi","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5029480176","display_name":"Rui Chen","orcid":"https://orcid.org/0000-0002-1641-8562"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Rui Chen","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5068477431","display_name":"Xia Hu","orcid":"https://orcid.org/0000-0003-2234-3226"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xia Hu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":8,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":{"value":0.08596159,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"13623","last_page":"13639"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10181","display_name":"Natural Language Processing Techniques","score":0.9228000044822693,"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":0.9228000044822693,"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/ground-truth","display_name":"Ground truth","score":0.7554395198822021},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.6810389757156372},{"id":"https://openalex.org/keywords/ranking","display_name":"Ranking (information retrieval)","score":0.6689918041229248},{"id":"https://openalex.org/keywords/language-model","display_name":"Language model","score":0.5043634176254272},{"id":"https://openalex.org/keywords/natural-language-processing","display_name":"Natural language processing","score":0.43934279680252075},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.4182817339897156}],"concepts":[{"id":"https://openalex.org/C146849305","wikidata":"https://www.wikidata.org/wiki/Q370766","display_name":"Ground truth","level":2,"score":0.7554395198822021},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6810389757156372},{"id":"https://openalex.org/C189430467","wikidata":"https://www.wikidata.org/wiki/Q7293293","display_name":"Ranking (information retrieval)","level":2,"score":0.6689918041229248},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.5043634176254272},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.43934279680252075},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4182817339897156}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2025.findings-acl.700","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-acl.700","pdf_url":"https://aclanthology.org/2025.findings-acl.700.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.700","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-acl.700","pdf_url":"https://aclanthology.org/2025.findings-acl.700.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":[{"id":"https://openalex.org/G5113848710","display_name":null,"funder_award_id":"IIS-2310260","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"}],"funders":[{"id":"https://openalex.org/F4320306076","display_name":"National Science Foundation","ror":"https://ror.org/021nxhr62"}],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4412888217.pdf","grobid_xml":"https://content.openalex.org/works/W4412888217.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W4295532600","https://openalex.org/W2063823869","https://openalex.org/W2047973478","https://openalex.org/W2067569035","https://openalex.org/W2090985514","https://openalex.org/W2188500270","https://openalex.org/W2303858293","https://openalex.org/W2915512527","https://openalex.org/W51364034","https://openalex.org/W3204019825"],"abstract_inverted_index":{"Large":[0],"Language":[1],"Models":[2],"(LLMs)":[3],"are":[4,20],"increasingly":[5],"adopted":[6],"across":[7,90],"real-world":[8],"applications,":[9],"yet":[10],"traditional":[11],"evaluations":[12],"rely":[13],"on":[14,31],"expensive,":[15],"domain-specific":[16],"ground-truth":[17,124],"labels":[18],"that":[19,95],"often":[21],"unavailable":[22],"or":[23],"infeasible.We":[24],"introduce":[25],"a":[26,53,56,80,111],"ground-truth-free":[27],"evaluation":[28],"framework":[29,97],"focused":[30],"reasoning":[32,92],"consistency":[33],"and":[34,60,68,107,114],"instruction":[35],"following,":[36],"shifting":[37],"the":[38,84,121],"emphasis":[39],"from":[40,83],"correctness-which":[41],"is":[42,127],"elusive":[43],"without":[44],"labels-to":[45],"transparent,":[46],"coherent,":[47],"evidence-based":[48],"reasoning.Each":[49],"model":[50],"response":[51],"includes":[52],"direct":[54],"answer,":[55],"structured":[57],"multi-step":[58],"explanation,":[59],"supporting":[61],"evidence,":[62],"all":[63],"assessed":[64],"via":[65],"semantic":[66],"similarity":[67],"output":[69],"adherence":[70],"checks.We":[71],"further":[72],"propose":[73],"TopK-ReRank,":[74],"which":[75],"refines":[76],"rankings":[77],"by":[78],"constructing":[79],"consensus":[81],"answer":[82],"most":[85],"reliable":[86],"models,":[87],"reducing":[88],"ambiguity":[89],"diverse":[91],"styles.Experiments":[93],"show":[94],"our":[96],"outperforms":[98],"existing":[99],"label-free":[100],"methods,":[101],"including":[102],"majority":[103],"voting,":[104],"triplet":[105],"ranking,":[106],"peer-review":[108],"approaches,":[109],"providing":[110],"more":[112],"interpretable":[113],"efficient":[115],"alternative":[116],"for":[117],"evaluating":[118],"LLMs":[119],"in":[120],"absence":[122],"of":[123],"labels.Our":[125],"code":[126],"available":[128],"at":[129],"https://github.com/MorrisZJ/":[130],"ReasonerRank.":[131]},"counts_by_year":[],"updated_date":"2026-06-11T09:08:48.828518","created_date":"2025-10-10T00:00:00"}