{"id":"https://openalex.org/W7166803135","doi":"https://doi.org/10.18653/v1/2026.findings-acl.1406","title":"Breaking the Evaluation Paradox: Evaluating High-Entropy Search with Computationally Irreducible Constraints","display_name":"Breaking the Evaluation Paradox: Evaluating High-Entropy Search with Computationally Irreducible Constraints","publication_year":2026,"publication_date":"2026-01-01","ids":{"openalex":"https://openalex.org/W7166803135","doi":"https://doi.org/10.18653/v1/2026.findings-acl.1406"},"language":null,"primary_location":{"id":"doi:10.18653/v1/2026.findings-acl.1406","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.1406","pdf_url":"https://aclanthology.org/2026.findings-acl.1406.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 2026","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2026.findings-acl.1406.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5139715865","display_name":"Juntao Wu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Juntao Wu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139782840","display_name":"Wei Wen","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wei Wen","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139759554","display_name":"Xianting Huang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xianting Huang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139751673","display_name":"Shuai Pang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Shuai Pang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5079762822","display_name":"Ruizhi Qiao","orcid":"https://orcid.org/0000-0002-3663-0149"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ruizhi Qiao","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139748077","display_name":"Xing Sun","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xing Sun","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5139846325","display_name":"Ke Wang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ke Wang","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":0.84189439,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"28209","last_page":"28218"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11596","display_name":"Constraint Satisfaction and Optimization","score":0.12470000237226486,"subfield":{"id":"https://openalex.org/subfields/1705","display_name":"Computer Networks and Communications"},"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/T11596","display_name":"Constraint Satisfaction and Optimization","score":0.12470000237226486,"subfield":{"id":"https://openalex.org/subfields/1705","display_name":"Computer Networks and Communications"},"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/T11106","display_name":"Data Management and Algorithms","score":0.10260000079870224,"subfield":{"id":"https://openalex.org/subfields/1711","display_name":"Signal Processing"},"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/T12292","display_name":"Graph Theory and Algorithms","score":0.07880000025033951,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"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/minification","display_name":"Minification","score":0.30140000581741333},{"id":"https://openalex.org/keywords/feature","display_name":"Feature (linguistics)","score":0.27869999408721924},{"id":"https://openalex.org/keywords/constraint","display_name":"Constraint (computer-aided design)","score":0.2745000123977661},{"id":"https://openalex.org/keywords/algebra-over-a-field","display_name":"Algebra over a field","score":0.2587999999523163},{"id":"https://openalex.org/keywords/representation","display_name":"Representation (politics)","score":0.25360000133514404}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5074999928474426},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.4648999869823456},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.3562000095844269},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.33180001378059387},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.3239000141620636},{"id":"https://openalex.org/C147764199","wikidata":"https://www.wikidata.org/wiki/Q6865248","display_name":"Minification","level":2,"score":0.30140000581741333},{"id":"https://openalex.org/C2776401178","wikidata":"https://www.wikidata.org/wiki/Q12050496","display_name":"Feature (linguistics)","level":2,"score":0.27869999408721924},{"id":"https://openalex.org/C2776036281","wikidata":"https://www.wikidata.org/wiki/Q48769818","display_name":"Constraint (computer-aided design)","level":2,"score":0.2745000123977661},{"id":"https://openalex.org/C126255220","wikidata":"https://www.wikidata.org/wiki/Q141495","display_name":"Mathematical optimization","level":1,"score":0.2615000009536743},{"id":"https://openalex.org/C136119220","wikidata":"https://www.wikidata.org/wiki/Q1000660","display_name":"Algebra over a field","level":2,"score":0.2587999999523163},{"id":"https://openalex.org/C2776359362","wikidata":"https://www.wikidata.org/wiki/Q2145286","display_name":"Representation (politics)","level":3,"score":0.25360000133514404},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.25200000405311584}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2026.findings-acl.1406","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.1406","pdf_url":"https://aclanthology.org/2026.findings-acl.1406.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 2026","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2026.findings-acl.1406","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.1406","pdf_url":"https://aclanthology.org/2026.findings-acl.1406.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 2026","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7166803135.pdf","grobid_xml":"https://content.openalex.org/works/W7166803135.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Evaluating":[0],"the":[1,90,114,156],"exhaustive":[2,82,135],"search":[3,83,128,147,158],"capabilities":[4],"of":[5,57,116,167],"large":[6],"language":[7],"models":[8,40],"(LLMs)":[9],"is":[10],"plagued":[11],"by":[12,88,183],"a":[13,95,110,163,189,200],"fundamental":[14],"paradox:":[15],"verifying":[16],"completeness":[17],"requires":[18],"complete":[19],"ground":[20,28,172],"truth,":[21],"yet":[22,211],"highentropy":[23],"enumeration":[24],"tasks":[25,129],"make":[26],"such":[27],"truth":[29,173],"impossible":[30,143],"for":[31,41,108,144,192,203],"humans":[32],"to":[33,37,98,134,140,149,153,207],"create.This":[34],"causes":[35],"benchmarks":[36],"systematically":[38],"penalize":[39],"outperforming":[42],"their":[43],"human":[44],"annotators.Despite":[45],"rapid":[46],"progress":[47],"in":[48],"web-search":[49],"and":[50,62,174],"deep":[51],"research":[52],"agents-which":[53],"now":[54],"issue":[55],"hundreds":[56],"queries,":[58],"traverse":[59,155],"diverse":[60],"sites,":[61],"synthesize":[63],"long":[64],"reports-evaluation":[65],"still":[66],"largely":[67],"relies":[68],"on":[69,113],"partially":[70],"annotated":[71],"answer":[72],"sets,":[73],"LLMbased":[74],"judges,":[75],"or":[76,146],"single-answer":[77],"questions":[78],"that":[79,130],"avoid":[80],"genuinely":[81,154],"scenarios.We":[84],"break":[85],"this":[86],"paradox":[87],"shifting":[89],"evaluation":[91],"paradigm":[92],"from":[93],"simulating":[94],"messy":[96],"reality":[97],"constructing":[99],"computationally":[100,117,132],"pure":[101],"challenges.We":[102],"introduce":[103],"VERITAS":[104],"(Verifiable":[105],"Traversal":[106],"Assessment":[107],"Search),":[109],"framework":[111],"built":[112],"principle":[115],"irreducible":[118],"constraints.By":[119],"introducing":[120],"novel,":[121],"non-optimizable":[122],"constraints,":[123],"we":[124],"create":[125],"verifiable,":[126],"sparse-answer":[127],"are":[131,138],"equivalent":[133],"enumeration.These":[136],"constraints":[137],"easy":[139],"verify":[141],"but":[142,198],"LLMs":[145],"engines":[148],"optimize,":[150],"forcing":[151],"agents":[152],"entire":[157],"space.VERITAS":[159],"can":[160],"automatically":[161],"generate":[162],"virtually":[164],"infinite":[165],"number":[166],"test":[168],"cases":[169],"with":[170,178],"perfect":[171],"precise":[175],"difficulty":[176],"control,":[177],"marginal":[179],"instance":[180],"cost":[181],"dominated":[182],"hash":[184],"computations.This":[185],"provides":[186],"not":[187],"only":[188],"robust":[190],"benchmark":[191],"evaluating":[193],"systematic":[194],"exploration":[195],"under":[196],"uncertainty":[197],"also":[199],"scalable":[201],"method":[202],"generating":[204],"training":[205],"data":[206],"improve":[208],"these":[209],"crucial,":[210],"underdeveloped,":[212],"capabilities.":[213]},"counts_by_year":[],"updated_date":"2026-07-29T14:22:42.915294","created_date":"2026-07-02T00:00:00"}