{"id":"https://openalex.org/W7140132082","doi":"https://doi.org/10.18653/v1/2026.eacl-long.258","title":"RoSE: Round-robin Synthetic Data Evaluation for Selecting LLM Generators without Human Test Sets","display_name":"RoSE: Round-robin Synthetic Data Evaluation for Selecting LLM Generators without Human Test Sets","publication_year":2026,"publication_date":"2026-01-01","ids":{"openalex":"https://openalex.org/W7140132082","doi":"https://doi.org/10.18653/v1/2026.eacl-long.258"},"language":null,"primary_location":{"id":"doi:10.18653/v1/2026.eacl-long.258","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.eacl-long.258","pdf_url":"https://aclanthology.org/2026.eacl-long.258.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 19th Conference of the European Chapter of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2026.eacl-long.258.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5081841143","display_name":"J\u00e1n \u010cegi\u0148","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jan Cegin","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5078389036","display_name":"Branislav Pecher","orcid":"https://orcid.org/0000-0003-0344-8620"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Branislav Pecher","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5082763244","display_name":"Ivan Srba","orcid":"https://orcid.org/0000-0003-3511-5337"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ivan Srba","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5043199710","display_name":"Jakub \u0160imko","orcid":"https://orcid.org/0000-0003-0239-4237"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jakub Simko","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":0.0,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":{"value":0.25715502,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"5530","last_page":"5545"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10201","display_name":"Speech Recognition and Synthesis","score":0.0421999990940094,"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/T10201","display_name":"Speech Recognition and Synthesis","score":0.0421999990940094,"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/T10743","display_name":"Software Testing and Debugging Techniques","score":0.03959999978542328,"subfield":{"id":"https://openalex.org/subfields/1712","display_name":"Software"},"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/T12125","display_name":"Aerospace and Aviation Technology","score":0.027400000020861626,"subfield":{"id":"https://openalex.org/subfields/2202","display_name":"Aerospace Engineering"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/synthetic-data","display_name":"Synthetic data","score":0.41600000858306885},{"id":"https://openalex.org/keywords/test","display_name":"Test (biology)","score":0.34779998660087585},{"id":"https://openalex.org/keywords/test-data","display_name":"Test data","score":0.32440000772476196},{"id":"https://openalex.org/keywords/selection","display_name":"Selection (genetic algorithm)","score":0.2973000109195709},{"id":"https://openalex.org/keywords/set","display_name":"Set (abstract data type)","score":0.29600000381469727}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5206000208854675},{"id":"https://openalex.org/C160920958","wikidata":"https://www.wikidata.org/wiki/Q7662746","display_name":"Synthetic data","level":2,"score":0.41600000858306885},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3700000047683716},{"id":"https://openalex.org/C2777267654","wikidata":"https://www.wikidata.org/wiki/Q3519023","display_name":"Test (biology)","level":2,"score":0.34779998660087585},{"id":"https://openalex.org/C16910744","wikidata":"https://www.wikidata.org/wiki/Q7705759","display_name":"Test data","level":2,"score":0.32440000772476196},{"id":"https://openalex.org/C81917197","wikidata":"https://www.wikidata.org/wiki/Q628760","display_name":"Selection (genetic algorithm)","level":2,"score":0.2973000109195709},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.2973000109195709},{"id":"https://openalex.org/C177264268","wikidata":"https://www.wikidata.org/wiki/Q1514741","display_name":"Set (abstract data type)","level":2,"score":0.29600000381469727},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.27959999442100525},{"id":"https://openalex.org/C2780992000","wikidata":"https://www.wikidata.org/wiki/Q17016113","display_name":"Generator (circuit theory)","level":3,"score":0.27649998664855957},{"id":"https://openalex.org/C58489278","wikidata":"https://www.wikidata.org/wiki/Q1172284","display_name":"Data set","level":2,"score":0.27410000562667847},{"id":"https://openalex.org/C115901376","wikidata":"https://www.wikidata.org/wiki/Q184199","display_name":"Automation","level":2,"score":0.26969999074935913}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2026.eacl-long.258","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.eacl-long.258","pdf_url":"https://aclanthology.org/2026.eacl-long.258.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 19th Conference of the European Chapter of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2026.eacl-long.258","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.eacl-long.258","pdf_url":"https://aclanthology.org/2026.eacl-long.258.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 19th Conference of the European Chapter of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[{"id":"https://openalex.org/G260610314","display_name":"Low Resource Artificial Intelligence","funder_award_id":"101136646","funder_id":"https://openalex.org/F4320320300","funder_display_name":"European Commission"},{"id":"https://openalex.org/G941802975","display_name":null,"funder_award_id":"90254","funder_id":"https://openalex.org/F4320320300","funder_display_name":"European Commission"}],"funders":[{"id":"https://openalex.org/F4320320300","display_name":"European Commission","ror":"https://ror.org/00k4n6c32"}],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7140132082.pdf","grobid_xml":"https://content.openalex.org/works/W7140132082.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"LLMs":[0,27,33],"are":[1,8,40,59],"powerful":[2],"generators":[3],"of":[4,99,124,159,169],"synthetic":[5,110],"data,":[6],"which":[7],"used":[9],"for":[10,18,41,62,82],"training":[11,174],"smaller,":[12],"specific":[13],"models.This":[14],"is":[15,24,49,120,165,194],"especially":[16],"valuable":[17],"low-resource":[19],"languages,":[20,131],"where":[21],"human-labelled":[22,190],"data":[23,76],"scarce":[25],"but":[26],"can":[28],"still":[29],"produce":[30],"highquality":[31],"text.However,":[32],"differ":[34],"in":[35,167],"how":[36],"useful":[37],"their":[38],"outputs":[39,98,182],"training.Selecting":[42],"the":[43,84,97,121,140,160,179,195],"best":[44,85],"LLM":[45,86],"as":[46],"a":[47,79,93,100,175,200],"generator":[48,87,102,142,162],"challenging":[50],"because":[51],"extrinsic":[52],"evaluation":[53],"requires":[54],"costly":[55],"human":[56,89,206],"annotations":[57],"(which":[58],"often":[60,144],"unavailable":[61],"lowresource":[63],"languages),":[64],"while":[65],"intrinsic":[66,148,151],"metrics":[67],"correlate":[68],"poorly":[69],"with":[70,203],"downstream":[71,170],"performance.We":[72],"introduce":[73],"Round-robin":[74],"Synthetic":[75],"Evaluation":[77],"(RoSE),":[78],"proxy":[80],"metric":[81,197],"selecting":[83],"without":[88],"test":[90,191,207],"sets.RoSE":[91],"trains":[92],"small":[94,126,176],"model":[95,177],"on":[96,108,178,189,205],"candidate":[101,115],"(LLM)":[103],"and":[104,132,153,186],"then":[105],"evaluates":[106],"it":[107,188],"generated":[109],"examples":[111],"from":[112],"all":[113],"other":[114,147],"LLMs.The":[116],"final":[117],"RoSE":[118,138,193],"score":[119],"mean":[122],"performance":[123,204],"this":[125],"model.Across":[127],"six":[128],"LLMs,":[129],"eleven":[130],"three":[133],"tasks":[134],"(sentiment,":[135],"topic,":[136],"intent),":[137],"identifies":[139],"optimal":[141,161],"more":[143],"than":[145],"any":[146],"heuristics.RoSE":[149],"outperforms":[150],"heuristics":[152],"comes":[154],"within":[155],"0.76":[156],"percentage":[157],"points":[158],"baseline.This":[163],"result":[164],"measured":[166],"terms":[168],"performance,":[171],"obtained":[172],"by":[173],"chosen":[180],"generator's":[181],"(optimal":[183],"vs.":[184],"proxy-metric-selected)":[185],"evaluating":[187],"data.Additionally,":[192],"only":[196],"to":[198],"achieve":[199],"positive":[201],"correlation":[202],"data.":[208]},"counts_by_year":[],"updated_date":"2026-07-29T14:22:42.915294","created_date":"2026-03-24T00:00:00"}