{"id":"https://openalex.org/W4414957384","doi":"https://doi.org/10.18653/v1/2026.acl-long.126","title":"SCAN: Structured Capability Assessment and Navigation for LLMs","display_name":"SCAN: Structured Capability Assessment and Navigation for LLMs","publication_year":2026,"publication_date":"2026-01-01","ids":{"openalex":"https://openalex.org/W4414957384","doi":"https://doi.org/10.18653/v1/2026.acl-long.126"},"language":"en","primary_location":{"id":"doi:10.18653/v1/2026.acl-long.126","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.acl-long.126","pdf_url":"https://aclanthology.org/2026.acl-long.126.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 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["arxiv","crossref","datacite"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2026.acl-long.126.pdf","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5033385054","display_name":"Z.B. Wang","orcid":"https://orcid.org/0009-0001-7812-9504"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zongqi Wang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5119902065","display_name":"Tianle Gu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Tianle Gu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100389824","display_name":"Chen Gong","orcid":"https://orcid.org/0000-0001-9305-6908"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chen Gong","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5024815756","display_name":"Xin Tian","orcid":"https://orcid.org/0000-0001-6402-7016"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xin Tian","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Siqi Bao","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Siqi Bao","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":null,"display_name":"Yujiu Yang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yujiu Yang","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":0.00544285,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"2760","last_page":"2799"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11986","display_name":"Scientific Computing and Data Management","score":0.9210000038146973,"subfield":{"id":"https://openalex.org/subfields/1802","display_name":"Information Systems and Management"},"field":{"id":"https://openalex.org/fields/18","display_name":"Decision Sciences"},"domain":{"id":"https://openalex.org/domains/2","display_name":"Social Sciences"}},"topics":[{"id":"https://openalex.org/T11986","display_name":"Scientific Computing and Data Management","score":0.9210000038146973,"subfield":{"id":"https://openalex.org/subfields/1802","display_name":"Information Systems and Management"},"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/suite","display_name":"Suite","score":0.6628000140190125},{"id":"https://openalex.org/keywords/key","display_name":"Key (lock)","score":0.5428000092506409},{"id":"https://openalex.org/keywords/visualization","display_name":"Visualization","score":0.48539999127388},{"id":"https://openalex.org/keywords/construct","display_name":"Construct (python library)","score":0.4848000109195709},{"id":"https://openalex.org/keywords/taxonomy","display_name":"Taxonomy (biology)","score":0.46790000796318054},{"id":"https://openalex.org/keywords/data-visualization","display_name":"Data visualization","score":0.3158999979496002}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7434999942779541},{"id":"https://openalex.org/C79581498","wikidata":"https://www.wikidata.org/wiki/Q1367530","display_name":"Suite","level":2,"score":0.6628000140190125},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.5428000092506409},{"id":"https://openalex.org/C36464697","wikidata":"https://www.wikidata.org/wiki/Q451553","display_name":"Visualization","level":2,"score":0.48539999127388},{"id":"https://openalex.org/C2780801425","wikidata":"https://www.wikidata.org/wiki/Q5164392","display_name":"Construct (python library)","level":2,"score":0.4848000109195709},{"id":"https://openalex.org/C58642233","wikidata":"https://www.wikidata.org/wiki/Q8269924","display_name":"Taxonomy (biology)","level":2,"score":0.46790000796318054},{"id":"https://openalex.org/C2522767166","wikidata":"https://www.wikidata.org/wiki/Q2374463","display_name":"Data science","level":1,"score":0.46560001373291016},{"id":"https://openalex.org/C115903868","wikidata":"https://www.wikidata.org/wiki/Q80993","display_name":"Software engineering","level":1,"score":0.4025999903678894},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.32839998602867126},{"id":"https://openalex.org/C172367668","wikidata":"https://www.wikidata.org/wiki/Q6504956","display_name":"Data visualization","level":3,"score":0.3158999979496002},{"id":"https://openalex.org/C168167062","wikidata":"https://www.wikidata.org/wiki/Q1117970","display_name":"Component (thermodynamics)","level":2,"score":0.2969000041484833},{"id":"https://openalex.org/C2777617010","wikidata":"https://www.wikidata.org/wiki/Q18957","display_name":"Mainstream","level":2,"score":0.2922999858856201},{"id":"https://openalex.org/C174348530","wikidata":"https://www.wikidata.org/wiki/Q188635","display_name":"Bridging (networking)","level":2,"score":0.26339998841285706},{"id":"https://openalex.org/C25343380","wikidata":"https://www.wikidata.org/wiki/Q277521","display_name":"Relation (database)","level":2,"score":0.2621999979019165}],"mesh":[],"locations_count":3,"locations":[{"id":"doi:10.18653/v1/2026.acl-long.126","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.acl-long.126","pdf_url":"https://aclanthology.org/2026.acl-long.126.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 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},{"id":"pmh:oai:arXiv.org:2505.06698","is_oa":true,"landing_page_url":"https://arxiv.org/abs/2505.06698","pdf_url":"https://arxiv.org/pdf/2505.06698","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2505.06698","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2505.06698","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.18653/v1/2026.acl-long.126","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.acl-long.126","pdf_url":"https://aclanthology.org/2026.acl-long.126.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 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[{"id":"https://openalex.org/G2275985752","display_name":null,"funder_award_id":"62576191","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"}],"funders":[{"id":"https://openalex.org/F4320321001","display_name":"National Natural Science Foundation of China","ror":"https://ror.org/01h0zpd94"}],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4414957384.pdf","grobid_xml":"https://content.openalex.org/works/W4414957384.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Evaluating":[0],"Large":[1],"Language":[2],"Models":[3],"(LLMs)":[4],"has":[5,22],"become":[6],"increasingly":[7],"important,":[8],"with":[9,36],"automatic":[10],"evaluation":[11,103,158,189],"benchmarks":[12,29],"gaining":[13],"prominence":[14],"as":[15],"alternatives":[16],"to":[17,31,86,149,177],"human":[18],"evaluation.While":[19],"existing":[20],"research":[21],"focused":[23],"on":[24],"approximating":[25],"model":[26,124],"rankings,":[27],"such":[28],"fail":[30],"provide":[32],"users":[33],"and":[34,39,56,70,97,114,121,126],"developers":[35],"a":[37,43,58,88,94,110,128,156],"comprehensive":[38,69,157],"fine-grained":[40,71,188],"understanding":[41,192],"of":[42,65,112,123,143,159,165,181,187],"specific":[44],"model's":[45],"capabilities.To":[46],"fill":[47],"this":[48],"gap,":[49],"we":[50,154],"propose":[51],"SCAN":[52],"(Structured":[53],"Capability":[54],"Assessment":[55],"Navigation),":[57],"practical":[59],"framework":[60],"that":[61,100,117,135],"enables":[62],"detailed":[63,163],"characterization":[64],"LLM":[66,193],"capabilities":[67],"through":[68],"evaluation.SCAN":[72],"incorporates":[73],"four":[74],"key":[75],"components:":[76],"(1)":[77],"TaxBuilder,":[78],"which":[79],"extracts":[80],"capability-indicating":[81],"tags":[82],"from":[83],"extensive":[84],"queries":[85],"construct":[87],"hierarchical":[89],"taxonomy":[90],"automatically;":[91],"(2)":[92],"RealMix,":[93],"query":[95],"synthesis":[96],"filtering":[98],"mechanism":[99],"ensures":[101],"sufficient":[102],"data":[104],"for":[105],"each":[106],"capability":[107],"tag;":[108],"(3)":[109],"suite":[111],"visualization":[113],"analysis":[115,122,164],"tools":[116],"facilitate":[118],"efficient":[119],"navigation":[120],"capabilities;":[125],"(4)":[127],"PC":[129],"2based":[130],"(Pre-Comparison-derived":[131],"Criteria)":[132],"LLMas-a-Judge":[133],"approach":[134],"achieves":[136],"significantly":[137],"higher":[138],"accuracy":[139,144],"(see":[140],"the":[141,166,178,185],"definition":[142],"in":[145,190],"\u00a7":[146],"D)":[147],"compared":[148],"classic":[150],"LLM-as-a-Judge":[151],"method.Using":[152],"SCAN,":[153],"conduct":[155],"21":[160],"mainstream":[161],"LLMs.Our":[162],"GPT-OSS":[167],"family":[168],"reveals":[169],"substantial":[170],"performance":[171],"variations,":[172],"even":[173],"within":[174],"sub-capabilities":[175],"belonging":[176],"same":[179],"category":[180],"capability.This":[182],"finding":[183],"highlights":[184],"importance":[186],"accurately":[191],"behavior.":[194]},"counts_by_year":[],"updated_date":"2026-08-05T07:39:15.569665","created_date":"2025-10-09T00:00:00"}