{"id":"https://openalex.org/W4412377897","doi":"https://doi.org/10.1145/3726302.3730221","title":"Limitations of Automatic Relevance Assessments with Large Language Models for Fair and Reliable Retrieval Evaluation","display_name":"Limitations of Automatic Relevance Assessments with Large Language Models for Fair and Reliable Retrieval Evaluation","publication_year":2025,"publication_date":"2025-07-13","ids":{"openalex":"https://openalex.org/W4412377897","doi":"https://doi.org/10.1145/3726302.3730221"},"language":"en","primary_location":{"id":"doi:10.1145/3726302.3730221","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3726302.3730221","pdf_url":"https://dl.acm.org/doi/pdf/10.1145/3726302.3730221","source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 48th International ACM SIGIR Conference on Research and Development in Information Retrieval","raw_type":"proceedings-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://dl.acm.org/doi/pdf/10.1145/3726302.3730221","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5070408128","display_name":"David Otero","orcid":"https://orcid.org/0000-0003-1139-0449"},"institutions":[{"id":"https://openalex.org/I11019714","display_name":"Universidade da Coru\u00f1a","ror":"https://ror.org/01qckj285","country_code":"ES","type":"education","lineage":["https://openalex.org/I11019714"]}],"countries":["ES"],"is_corresponding":false,"raw_author_name":"David Otero","raw_affiliation_strings":["IRlab, CITIC, Universidade da Coru\u00f1a, A Coru\u00f1a, Spain"],"raw_orcid":"https://orcid.org/0000-0003-1139-0449","affiliations":[{"raw_affiliation_string":"IRlab, CITIC, Universidade da Coru\u00f1a, A Coru\u00f1a, Spain","institution_ids":["https://openalex.org/I11019714"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5046723532","display_name":"Javier Parapar","orcid":"https://orcid.org/0000-0002-5997-8252"},"institutions":[{"id":"https://openalex.org/I11019714","display_name":"Universidade da Coru\u00f1a","ror":"https://ror.org/01qckj285","country_code":"ES","type":"education","lineage":["https://openalex.org/I11019714"]}],"countries":["ES"],"is_corresponding":false,"raw_author_name":"Javier Parapar","raw_affiliation_strings":["IRlab, CITIC, Universidade da Coru\u00f1a, A Coru\u00f1a, Spain"],"raw_orcid":"https://orcid.org/0000-0002-5997-8252","affiliations":[{"raw_affiliation_string":"IRlab, CITIC, Universidade da Coru\u00f1a, A Coru\u00f1a, Spain","institution_ids":["https://openalex.org/I11019714"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5056469959","display_name":"\u00c1lvaro Barreiro","orcid":"https://orcid.org/0000-0002-6698-2946"},"institutions":[{"id":"https://openalex.org/I11019714","display_name":"Universidade da Coru\u00f1a","ror":"https://ror.org/01qckj285","country_code":"ES","type":"education","lineage":["https://openalex.org/I11019714"]}],"countries":["ES"],"is_corresponding":false,"raw_author_name":"\u00c1lvaro Barreiro","raw_affiliation_strings":["IRlab, CITIC, University of A Coru\u00f1a, A Coru\u00f1a, Spain"],"raw_orcid":"https://orcid.org/0000-0002-6698-2946","affiliations":[{"raw_affiliation_string":"IRlab, CITIC, University of A Coru\u00f1a, A Coru\u00f1a, Spain","institution_ids":["https://openalex.org/I11019714"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I11019714"],"apc_list":null,"apc_paid":null,"fwci":5.584,"has_fulltext":false,"cited_by_count":4,"citation_normalized_percentile":{"value":0.95606818,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"cited_by_percentile_year":{"min":91,"max":99},"biblio":{"volume":null,"issue":null,"first_page":"2545","last_page":"2549"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.9983000159263611,"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/T10028","display_name":"Topic Modeling","score":0.9983000159263611,"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/T10286","display_name":"Information Retrieval and Search Behavior","score":0.9933000206947327,"subfield":{"id":"https://openalex.org/subfields/1710","display_name":"Information Systems"},"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/T11719","display_name":"Data Quality and Management","score":0.9869999885559082,"subfield":{"id":"https://openalex.org/subfields/1803","display_name":"Management Science and Operations Research"},"field":{"id":"https://openalex.org/fields/18","display_name":"Decision Sciences"},"domain":{"id":"https://openalex.org/domains/2","display_name":"Social Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/relevance","display_name":"Relevance (law)","score":0.7765898704528809},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.7551296353340149},{"id":"https://openalex.org/keywords/information-retrieval","display_name":"Information retrieval","score":0.529376745223999},{"id":"https://openalex.org/keywords/natural-language-processing","display_name":"Natural language processing","score":0.445195734500885},{"id":"https://openalex.org/keywords/relevance-feedback","display_name":"Relevance feedback","score":0.4304887056350708},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.3794408440589905},{"id":"https://openalex.org/keywords/image-retrieval","display_name":"Image retrieval","score":0.18986323475837708}],"concepts":[{"id":"https://openalex.org/C158154518","wikidata":"https://www.wikidata.org/wiki/Q7310970","display_name":"Relevance (law)","level":2,"score":0.7765898704528809},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7551296353340149},{"id":"https://openalex.org/C23123220","wikidata":"https://www.wikidata.org/wiki/Q816826","display_name":"Information retrieval","level":1,"score":0.529376745223999},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.445195734500885},{"id":"https://openalex.org/C2779532271","wikidata":"https://www.wikidata.org/wiki/Q445558","display_name":"Relevance feedback","level":4,"score":0.4304887056350708},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3794408440589905},{"id":"https://openalex.org/C1667742","wikidata":"https://www.wikidata.org/wiki/Q10927554","display_name":"Image retrieval","level":3,"score":0.18986323475837708},{"id":"https://openalex.org/C199539241","wikidata":"https://www.wikidata.org/wiki/Q7748","display_name":"Law","level":1,"score":0.0},{"id":"https://openalex.org/C115961682","wikidata":"https://www.wikidata.org/wiki/Q860623","display_name":"Image (mathematics)","level":2,"score":0.0},{"id":"https://openalex.org/C17744445","wikidata":"https://www.wikidata.org/wiki/Q36442","display_name":"Political science","level":0,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1145/3726302.3730221","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3726302.3730221","pdf_url":"https://dl.acm.org/doi/pdf/10.1145/3726302.3730221","source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 48th International ACM SIGIR Conference on Research and Development in Information Retrieval","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.1145/3726302.3730221","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3726302.3730221","pdf_url":"https://dl.acm.org/doi/pdf/10.1145/3726302.3730221","source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 48th International ACM SIGIR Conference on Research and Development in Information Retrieval","raw_type":"proceedings-article"},"sustainable_development_goals":[{"display_name":"Quality Education","score":0.49000000953674316,"id":"https://metadata.un.org/sdg/4"}],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4412377897.pdf"},"referenced_works_count":19,"referenced_works":["https://openalex.org/W1966835268","https://openalex.org/W2015338694","https://openalex.org/W2021581601","https://openalex.org/W2021856948","https://openalex.org/W2023217403","https://openalex.org/W2802553787","https://openalex.org/W2892923641","https://openalex.org/W2895814542","https://openalex.org/W2897412906","https://openalex.org/W3158689111","https://openalex.org/W4304686742","https://openalex.org/W4378533174","https://openalex.org/W4384107234","https://openalex.org/W4385688511","https://openalex.org/W4386043577","https://openalex.org/W4396822183","https://openalex.org/W4400526284","https://openalex.org/W4400526908","https://openalex.org/W4405143965"],"related_works":["https://openalex.org/W1921936017","https://openalex.org/W2001985945","https://openalex.org/W2009716188","https://openalex.org/W1518380457","https://openalex.org/W1971071004","https://openalex.org/W1973132420","https://openalex.org/W2460037195","https://openalex.org/W2142731558","https://openalex.org/W2077213532","https://openalex.org/W2157408137"],"abstract_inverted_index":{"Offline":[0],"evaluation":[1,141],"of":[2,17,38,164],"search":[3],"systems":[4,74,113,133],"depends":[5],"on":[6,95],"test":[7,32],"collections.":[8],"These":[9,80],"benchmarks":[10],"provide":[11],"the":[12,108],"researchers":[13],"with":[14,77,114],"a":[15],"corpus":[16],"documents,":[18],"topics":[19],"and":[20,103,134],"relevance":[21,63],"judgements":[22,106,127,150],"indicating":[23],"which":[24],"documents":[25],"are":[26,34,55,82,151],"relevant":[27],"for":[28,61],"each":[29],"topic.":[30],"While":[31],"collections":[33],"an":[35,160],"integral":[36],"part":[37],"Information":[39],"Retrieval":[40],"(IR)":[41],"research,":[42],"their":[43],"creation":[44],"involves":[45],"significant":[46,110],"efforts":[47],"in":[48,84],"manual":[49],"annotation.":[50],"Large":[51],"language":[52],"models":[53],"(LLMs)":[54],"gaining":[56],"much":[57],"attention":[58],"as":[59,142],"tools":[60],"automatic":[62],"assessment.":[64],"Recent":[65],"research":[66],"has":[67],"shown":[68],"that":[69,148],"LLM-based":[70,105,149],"assessments":[71],"yield":[72],"high":[73,162],"ranking":[75,129,154],"correlation":[76],"human-made":[78],"judgements.":[79,144],"correlations":[81,100],"helpful":[83],"large-scale":[85],"experiments":[86],"but":[87],"less":[88],"informative":[89],"if":[90],"we":[91,122,158],"want":[92],"to":[93,116],"focus":[94],"top-performing":[96,132,155],"systems.":[97,156],"Moreover,":[98,157],"these":[99],"ignore":[101],"whether":[102],"how":[104,125,136],"impact":[107],"statistically":[109],"differences":[111,130],"among":[112,131],"respect":[115],"human":[117,143],"assessments.":[118],"In":[119],"this":[120],"work,":[121],"look":[123],"at":[124,153],"LLM-generated":[126],"preserve":[128,138],"also":[135],"they":[137],"pairwise":[139],"significance":[140],"Our":[145],"results":[146],"show":[147],"unfair":[152],"observe":[159],"exceedingly":[161],"rate":[163],"false":[165],"positives":[166],"regarding":[167],"statistical":[168],"differences.":[169]},"counts_by_year":[{"year":2026,"cited_by_count":3},{"year":2025,"cited_by_count":1}],"updated_date":"2025-11-06T03:46:38.306776","created_date":"2025-10-10T00:00:00"}