{"id":"https://openalex.org/W4409361595","doi":"https://doi.org/10.1609/aaai.v39i25.34857","title":"ACPBench: Reasoning About Action, Change, and Planning","display_name":"ACPBench: Reasoning About Action, Change, and Planning","publication_year":2025,"publication_date":"2025-04-11","ids":{"openalex":"https://openalex.org/W4409361595","doi":"https://doi.org/10.1609/aaai.v39i25.34857"},"language":"en","primary_location":{"id":"doi:10.1609/aaai.v39i25.34857","is_oa":true,"landing_page_url":"https://doi.org/10.1609/aaai.v39i25.34857","pdf_url":"https://ojs.aaai.org/index.php/AAAI/article/download/34857/37012","source":{"id":"https://openalex.org/S4210191458","display_name":"Proceedings of the AAAI Conference on Artificial Intelligence","issn_l":"2159-5399","issn":["2159-5399","2374-3468"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/P4310320058","host_organization_name":"Association for the Advancement of Artificial Intelligence","host_organization_lineage":["https://openalex.org/P4310320058"],"host_organization_lineage_names":["Association for the Advancement of Artificial Intelligence"],"type":"conference"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the AAAI Conference on Artificial Intelligence","raw_type":"journal-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"diamond","oa_url":"https://ojs.aaai.org/index.php/AAAI/article/download/34857/37012","any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5025268861","display_name":"Harsha Kokel","orcid":"https://orcid.org/0000-0002-7548-3719"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Harsha Kokel","raw_affiliation_strings":["IBM Research"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"IBM Research","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5103010384","display_name":"Michael B. Katz","orcid":"https://orcid.org/0000-0003-1021-1402"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Michael Katz","raw_affiliation_strings":["IBM Research"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"IBM Research","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5085594669","display_name":"Kavitha Srinivas","orcid":"https://orcid.org/0000-0003-4610-967X"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Kavitha Srinivas","raw_affiliation_strings":["IBM Research"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"IBM Research","institution_ids":[]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5037217032","display_name":"Shirin Sohrabi","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Shirin Sohrabi","raw_affiliation_strings":["IBM Research"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"IBM Research","institution_ids":[]}]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":1.2703,"has_fulltext":true,"cited_by_count":1,"citation_normalized_percentile":{"value":0.71597914,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":91,"max":95},"biblio":{"volume":"39","issue":"25","first_page":"26559","last_page":"26568"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11450","display_name":"Model-Driven Software Engineering Techniques","score":0.14740000665187836,"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"}},"topics":[{"id":"https://openalex.org/T11450","display_name":"Model-Driven Software Engineering Techniques","score":0.14740000665187836,"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/T10430","display_name":"Software Engineering Techniques and Practices","score":0.13920000195503235,"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/T10906","display_name":"AI-based Problem Solving and Planning","score":0.1348000019788742,"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"}}],"keywords":[{"id":"https://openalex.org/keywords/action","display_name":"Action (physics)","score":0.6919803023338318},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.38598203659057617},{"id":"https://openalex.org/keywords/cognitive-science","display_name":"Cognitive science","score":0.37957534193992615},{"id":"https://openalex.org/keywords/psychology","display_name":"Psychology","score":0.34078335762023926},{"id":"https://openalex.org/keywords/process-management","display_name":"Process management","score":0.32464635372161865},{"id":"https://openalex.org/keywords/business","display_name":"Business","score":0.20965230464935303}],"concepts":[{"id":"https://openalex.org/C2780791683","wikidata":"https://www.wikidata.org/wiki/Q846785","display_name":"Action (physics)","level":2,"score":0.6919803023338318},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.38598203659057617},{"id":"https://openalex.org/C188147891","wikidata":"https://www.wikidata.org/wiki/Q147638","display_name":"Cognitive science","level":1,"score":0.37957534193992615},{"id":"https://openalex.org/C15744967","wikidata":"https://www.wikidata.org/wiki/Q9418","display_name":"Psychology","level":0,"score":0.34078335762023926},{"id":"https://openalex.org/C195094911","wikidata":"https://www.wikidata.org/wiki/Q14167904","display_name":"Process management","level":1,"score":0.32464635372161865},{"id":"https://openalex.org/C144133560","wikidata":"https://www.wikidata.org/wiki/Q4830453","display_name":"Business","level":0,"score":0.20965230464935303},{"id":"https://openalex.org/C62520636","wikidata":"https://www.wikidata.org/wiki/Q944","display_name":"Quantum mechanics","level":1,"score":0.0},{"id":"https://openalex.org/C121332964","wikidata":"https://www.wikidata.org/wiki/Q413","display_name":"Physics","level":0,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1609/aaai.v39i25.34857","is_oa":true,"landing_page_url":"https://doi.org/10.1609/aaai.v39i25.34857","pdf_url":"https://ojs.aaai.org/index.php/AAAI/article/download/34857/37012","source":{"id":"https://openalex.org/S4210191458","display_name":"Proceedings of the AAAI Conference on Artificial Intelligence","issn_l":"2159-5399","issn":["2159-5399","2374-3468"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/P4310320058","host_organization_name":"Association for the Advancement of Artificial Intelligence","host_organization_lineage":["https://openalex.org/P4310320058"],"host_organization_lineage_names":["Association for the Advancement of Artificial Intelligence"],"type":"conference"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the AAAI Conference on Artificial Intelligence","raw_type":"journal-article"}],"best_oa_location":{"id":"doi:10.1609/aaai.v39i25.34857","is_oa":true,"landing_page_url":"https://doi.org/10.1609/aaai.v39i25.34857","pdf_url":"https://ojs.aaai.org/index.php/AAAI/article/download/34857/37012","source":{"id":"https://openalex.org/S4210191458","display_name":"Proceedings of the AAAI Conference on Artificial Intelligence","issn_l":"2159-5399","issn":["2159-5399","2374-3468"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/P4310320058","host_organization_name":"Association for the Advancement of Artificial Intelligence","host_organization_lineage":["https://openalex.org/P4310320058"],"host_organization_lineage_names":["Association for the Advancement of Artificial Intelligence"],"type":"conference"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the AAAI Conference on Artificial Intelligence","raw_type":"journal-article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4409361595.pdf","grobid_xml":"https://content.openalex.org/works/W4409361595.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W4391375266","https://openalex.org/W2899084033","https://openalex.org/W2748952813","https://openalex.org/W2390279801","https://openalex.org/W4391913857","https://openalex.org/W2358668433","https://openalex.org/W4396701345","https://openalex.org/W2376932109","https://openalex.org/W2001405890","https://openalex.org/W4396696052"],"abstract_inverted_index":{"There":[0],"is":[1,32,74,163],"an":[2],"increasing":[3],"body":[4],"of":[5,59,64,105,122,138],"work":[6],"using":[7],"Large":[8],"Language":[9],"Models":[10],"(LLMs)":[11],"as":[12],"agents":[13],"for":[14,41,51],"orchestrating":[15],"workflows":[16],"and":[17,25,97,125],"making":[18],"decisions":[19],"in":[20,56,80,134,153],"domains":[21,78],"that":[22],"require":[23],"planning":[24,70,77],"multistep":[26],"reasoning.":[27],"As":[28],"a":[29,49,81,146],"result,":[30],"it":[31,100],"imperative":[33],"to":[34,87],"evaluate":[35],"LLMs":[36,124],"on":[37,155,165],"core":[38],"skills":[39],"required":[40],"planning.":[42,60],"In":[43],"this":[44],"work,":[45],"we":[46],"present":[47],"ACPBench,":[48],"benchmark":[50,62],"evaluating":[52],"the":[53,57,103,131,135,139],"reasoning":[54,66,128,136,148],"tasks":[55,67,96],"field":[58],"The":[61,72],"consists":[63],"7":[65],"over":[68],"13":[69],"domains.":[71,98],"collection":[73],"constructed":[75],"from":[76],"described":[79],"formal":[82],"language.":[83],"This":[84],"allows":[85,101],"us":[86,102],"synthesize":[88],"problems":[89,114],"with":[90,143],"provably":[91],"correct":[92],"solutions":[93],"across":[94],"many":[95],"Further,":[99],"luxury":[104],"scale":[106],"without":[107],"additional":[108,113],"human":[109],"effort,":[110],"i.e.,":[111],"many]":[112],"can":[115],"be":[116],"created":[117],"automatically.":[118],"Our":[119,141],"extensive":[120],"evaluation":[121],"21":[123],"OpenAI":[126,144],"o1":[127],"models":[129],"highlight":[130],"significant":[132,151],"gap":[133],"capability":[137],"LLMs.":[140],"findings":[142],"o1,":[145],"multi-turn":[147],"model,":[149],"reveal":[150],"gains":[152],"performance":[154],"multiple-choice":[156],"questions,":[157],"yet":[158],"surprisingly,":[159],"no":[160],"notable":[161],"progress":[162],"made":[164],"boolean":[166],"questions.":[167]},"counts_by_year":[{"year":2025,"cited_by_count":1}],"updated_date":"2026-07-29T14:22:42.915294","created_date":"2025-10-10T00:00:00"}