{"id":"https://openalex.org/W7207598790","doi":"https://doi.org/10.26434/chemrxiv.15008241/v1","title":"Systematic evaluation of LLM-based zero-shot model construction in chemistry and materials science: predictive performance and response reliability","display_name":"Systematic evaluation of LLM-based zero-shot model construction in chemistry and materials science: predictive performance and response reliability","publication_year":2026,"publication_date":"2026-09-03","ids":{"openalex":"https://openalex.org/W7207598790","doi":"https://doi.org/10.26434/chemrxiv.15008241/v1"},"language":"en","primary_location":{"id":"doi:10.26434/chemrxiv.15008241/v1","doi":"https://doi.org/10.26434/chemrxiv.15008241/v1","is_oa":true,"landing_page_url":"https://doi.org/10.26434/chemrxiv.15008241/v1","pdf_url":"https://chemrxiv.org/doi/pdf/10.26434/chemrxiv.15008241/v1","source":{"id":"https://openalex.org/S4393918830","display_name":"ChemRxiv","issn_l":"2573-2293","issn":["2573-2293"],"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"acceptedVersion","is_accepted":true,"is_published":false,"raw_source_name":null,"raw_type":"posted-content"},"volume":null,"type":"preprint","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://chemrxiv.org/doi/pdf/10.26434/chemrxiv.15008241/v1","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5149458268","display_name":"Shunya Maezawa","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I206011266","display_name":"Kwansei Gakuin University","ror":"https://ror.org/02qf2tx24","country_code":"JP","type":"education","lineage":["https://openalex.org/I206011266"]}],"countries":["JP"],"is_corresponding":false,"raw_author_name":"Shunya Maezawa","raw_affiliation_strings":["\ud835\udc4e Institute of Business and Accounting , Kwansei Gakuin University , 1-155 Uegahara Ichiban-cho , Nishinomiya , Hyogo 662-8501 , Japan"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"\ud835\udc4e Institute of Business and Accounting , Kwansei Gakuin University , 1-155 Uegahara Ichiban-cho , Nishinomiya , Hyogo 662-8501 , Japan","institution_ids":["https://openalex.org/I206011266"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5144253946","display_name":"Tsuyoshi Ueno","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I206011266","display_name":"Kwansei Gakuin University","ror":"https://ror.org/02qf2tx24","country_code":"JP","type":"education","lineage":["https://openalex.org/I206011266"]}],"countries":["JP"],"is_corresponding":false,"raw_author_name":"Tsuyoshi Ueno","raw_affiliation_strings":["\ud835\udc4e Institute of Business and Accounting , Kwansei Gakuin University , 1-155 Uegahara Ichiban-cho , Nishinomiya , Hyogo 662-8501 , Japan"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"\ud835\udc4e Institute of Business and Accounting , Kwansei Gakuin University , 1-155 Uegahara Ichiban-cho , Nishinomiya , Hyogo 662-8501 , Japan","institution_ids":["https://openalex.org/I206011266"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I206011266"],"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":null,"last_page":null},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11948","display_name":"Machine Learning in Materials Science","score":1.0,"subfield":{"id":"https://openalex.org/subfields/2505","display_name":"Materials Chemistry"},"field":{"id":"https://openalex.org/fields/25","display_name":"Materials Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},"topics":[{"id":"https://openalex.org/T11948","display_name":"Machine Learning in Materials Science","score":1.0,"subfield":{"id":"https://openalex.org/subfields/2505","display_name":"Materials Chemistry"},"field":{"id":"https://openalex.org/fields/25","display_name":"Materials Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10211","display_name":"Computational Drug Discovery Methods","score":0.0,"subfield":{"id":"https://openalex.org/subfields/1703","display_name":"Computational Theory and Mathematics"},"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/T10203","display_name":"Recommender Systems and Techniques","score":0.0,"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"}}],"keywords":[{"id":"https://openalex.org/keywords/large-language-models","display_name":"large language models","score":0.972000002861023},{"id":"https://openalex.org/keywords/materials-discovery","display_name":"materials discovery","score":0.5799999833106995},{"id":"https://openalex.org/keywords/preference-learning","display_name":"preference learning","score":0.8759999871253967},{"id":"https://openalex.org/keywords/pairwise-comparison","display_name":"pairwise comparison","score":0.8040000200271606},{"id":"https://openalex.org/keywords/position-bias","display_name":"position bias","score":0.7450000047683716},{"id":"https://openalex.org/keywords/preference-consistency","display_name":"preference consistency","score":0.7289999723434448},{"id":"https://openalex.org/keywords/ensemble-methods","display_name":"ensemble methods","score":0.42899999022483826},{"id":"https://openalex.org/keywords/response-reliability","display_name":"response reliability","score":0.5799999833106995}],"concepts":[{"id":"https://openalex.org/C184898388","wikidata":"https://www.wikidata.org/wiki/Q1435712","display_name":"Pairwise comparison","level":2,"score":0.6924999952316284},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6800000071525574},{"id":"https://openalex.org/C153083717","wikidata":"https://www.wikidata.org/wiki/Q6535263","display_name":"Leverage (statistics)","level":2,"score":0.6179999709129333},{"id":"https://openalex.org/C43214815","wikidata":"https://www.wikidata.org/wiki/Q7310987","display_name":"Reliability (semiconductor)","level":3,"score":0.5748000144958496},{"id":"https://openalex.org/C2776436953","wikidata":"https://www.wikidata.org/wiki/Q5163215","display_name":"Consistency (knowledge bases)","level":2,"score":0.5649999976158142},{"id":"https://openalex.org/C2781249084","wikidata":"https://www.wikidata.org/wiki/Q908656","display_name":"Preference","level":2,"score":0.5526999831199646},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.5414000153541565},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.535099983215332},{"id":"https://openalex.org/C164226766","wikidata":"https://www.wikidata.org/wiki/Q7293202","display_name":"Rank (graph theory)","level":2,"score":0.44690001010894775},{"id":"https://openalex.org/C181204326","wikidata":"https://www.wikidata.org/wiki/Q7239820","display_name":"Preference learning","level":3,"score":0.436599999666214},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.43230000138282776},{"id":"https://openalex.org/C2780801425","wikidata":"https://www.wikidata.org/wiki/Q5164392","display_name":"Construct (python library)","level":2,"score":0.4212999939918518},{"id":"https://openalex.org/C2778136018","wikidata":"https://www.wikidata.org/wiki/Q10350689","display_name":"Predictive power","level":2,"score":0.3603000044822693},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.3181000053882599},{"id":"https://openalex.org/C917703","wikidata":"https://www.wikidata.org/wiki/Q7239668","display_name":"Predictive inference","level":5,"score":0.31690001487731934},{"id":"https://openalex.org/C198082294","wikidata":"https://www.wikidata.org/wiki/Q3399648","display_name":"Position (finance)","level":2,"score":0.28610000014305115},{"id":"https://openalex.org/C49937458","wikidata":"https://www.wikidata.org/wiki/Q2599292","display_name":"Probabilistic logic","level":2,"score":0.2685000002384186},{"id":"https://openalex.org/C189430467","wikidata":"https://www.wikidata.org/wiki/Q7293293","display_name":"Ranking (information retrieval)","level":2,"score":0.2524999976158142},{"id":"https://openalex.org/C189950617","wikidata":"https://www.wikidata.org/wiki/Q937228","display_name":"Property (philosophy)","level":2,"score":0.25110000371932983}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.26434/chemrxiv.15008241/v1","doi":"https://doi.org/10.26434/chemrxiv.15008241/v1","is_oa":true,"landing_page_url":"https://doi.org/10.26434/chemrxiv.15008241/v1","pdf_url":"https://chemrxiv.org/doi/pdf/10.26434/chemrxiv.15008241/v1","source":{"id":"https://openalex.org/S4393918830","display_name":"ChemRxiv","issn_l":"2573-2293","issn":["2573-2293"],"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"acceptedVersion","is_accepted":true,"is_published":false,"raw_source_name":null,"raw_type":"posted-content"}],"best_oa_location":{"id":"doi:10.26434/chemrxiv.15008241/v1","doi":"https://doi.org/10.26434/chemrxiv.15008241/v1","is_oa":true,"landing_page_url":"https://doi.org/10.26434/chemrxiv.15008241/v1","pdf_url":"https://chemrxiv.org/doi/pdf/10.26434/chemrxiv.15008241/v1","source":{"id":"https://openalex.org/S4393918830","display_name":"ChemRxiv","issn_l":"2573-2293","issn":["2573-2293"],"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"acceptedVersion","is_accepted":true,"is_published":false,"raw_source_name":null,"raw_type":"posted-content"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[{"id":"https://openalex.org/F4320334764","display_name":"Japan Society for the Promotion of Science","ror":"https://ror.org/00hhkn466"}],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7207598790.pdf","grobid_xml":"https://content.openalex.org/works/W7207598790.grobid-xml"},"referenced_works_count":8,"referenced_works":["https://openalex.org/W4285429195","https://openalex.org/W4377023704","https://openalex.org/W4399453666","https://openalex.org/W4399695623","https://openalex.org/W4399912177","https://openalex.org/W4404088575","https://openalex.org/W4406892397","https://openalex.org/W7129785866"],"related_works":["https://openalex.org/W4410830098","https://openalex.org/W4410022944","https://openalex.org/W7213934344","https://openalex.org/W7221726070","https://openalex.org/W7162803379","https://openalex.org/W4417092504","https://openalex.org/W7154306896","https://openalex.org/W7133298251","https://openalex.org/W4413820841","https://openalex.org/W4412598279"],"abstract_inverted_index":{"Purely":[0],"data-driven":[1],"materials":[2,78],"exploration":[3,12],"is":[4],"often":[5],"constrained":[6],"by":[7,139],"limited":[8],"experimental":[9,48],"data,":[10],"motivating":[11],"strategies":[13],"that":[14,34,46],"leverage":[15],"prior":[16],"knowledge":[17],"of":[18,65,93,104,124,180,187,199],"the":[19,63,118,122,154,160,177,185,188,195],"target":[20,54],"material":[21],"system":[22,64],"encoded":[23],"in":[24,91,102],"large":[25],"language":[26],"models":[27,45],"(LLMs).":[28],"This":[29],"paper":[30],"describes":[31],"an":[32],"approach":[33],"combines":[35],"LLM-derived":[36],"pairwise":[37],"comparisons":[38],"with":[39,50,163],"preference":[40,107,127,144],"learning":[41,128],"to":[42,52],"construct":[43],"utility-function":[44],"rank":[47,97],"conditions":[49],"respect":[51],"a":[53],"property,":[55],"without":[56],"directly":[57],"using":[58],"measured":[59],"target-property":[60,131],"values":[61],"for":[62],"interest":[66],"either":[67],"during":[68],"LLM":[69,190],"inference":[70],"or":[71],"model":[72],"fitting.":[73],"Across":[74],"nine":[75],"chemistry":[76],"and":[77,81,96,99,109,126,143,165,184,197],"science":[79],"datasets":[80],"six":[82],"LLMs":[83],"from":[84],"two":[85],"providers,":[86],"we":[87],"evaluated":[88],"predictive":[89,178],"performance":[90,113,179],"terms":[92,103],"pairwise-comparison":[94],"accuracy":[95],"agreement,":[98],"response":[100],"reliability":[101,186],"position":[105],"bias,":[106],"consistency,":[108],"abstention":[110,152],"behavior.":[111],"Predictive":[112],"depended":[114],"more":[115,133],"strongly":[116],"on":[117,121],"dataset":[119],"than":[120,135],"choice":[123],"LLM,":[125],"generally":[129],"reproduced":[130],"rankings":[132],"accurately":[134],"direct":[136],"property":[137],"prediction":[138],"LLMs.":[140],"Position":[141],"bias":[142],"consistency":[145],"varied":[146],"across":[147,159],"dataset\u2013LLM":[148],"combinations.":[149],"Among":[150],"11":[151],"configurations,":[153],"ensemble":[155],"performed":[156],"best":[157],"overall":[158],"risk\u2013coverage":[161],"metrics,":[162],"consistency-based":[164],"LLM-judge-based":[166],"configurations":[167],"also":[168],"performing":[169],"relatively":[170],"well.":[171],"Our":[172],"systematic":[173],"examination":[174],"has":[175],"characterized":[176],"LLM-based":[181],"zero-shot":[182],"modeling":[183],"underlying":[189],"output,":[191],"providing":[192],"insights":[193],"into":[194],"potential":[196],"challenges":[198],"this":[200],"approach.":[201]},"counts_by_year":[],"updated_date":"2026-11-04T10:54:05.112534","created_date":"2026-09-04T00:00:00"}