{"id":"https://openalex.org/W4412889711","doi":"https://doi.org/10.18653/v1/2025.acl-long.1499","title":"PlanningArena: A Modular Benchmark for Multidimensional Evaluation of Planning and Tool Learning","display_name":"PlanningArena: A Modular Benchmark for Multidimensional Evaluation of Planning and Tool Learning","publication_year":2025,"publication_date":"2025-01-01","ids":{"openalex":"https://openalex.org/W4412889711","doi":"https://doi.org/10.18653/v1/2025.acl-long.1499"},"language":"en","primary_location":{"id":"doi:10.18653/v1/2025.acl-long.1499","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.acl-long.1499","pdf_url":"https://aclanthology.org/2025.acl-long.1499.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 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2025.acl-long.1499.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5084171262","display_name":"Zihan Zheng","orcid":"https://orcid.org/0000-0003-1486-9586"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zihan Zheng","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Tianle Cui","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Tianle Cui","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Chuwen Xie","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chuwen Xie","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5023225646","display_name":"Jiahui Pan","orcid":"https://orcid.org/0000-0002-7576-6743"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jiahui Pan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5086038509","display_name":"Qianglong Chen","orcid":"https://orcid.org/0000-0002-7845-1544"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Qianglong Chen","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5112502954","display_name":"Lewei He","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Lewei He","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":6,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":1.7588,"has_fulltext":true,"cited_by_count":1,"citation_normalized_percentile":{"value":0.8780149,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":95,"max":98},"biblio":{"volume":null,"issue":null,"first_page":"31047","last_page":"31086"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11902","display_name":"Intelligent Tutoring Systems and Adaptive Learning","score":0.6001999974250793,"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/T11902","display_name":"Intelligent Tutoring Systems and Adaptive Learning","score":0.6001999974250793,"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/T11299","display_name":"Problem and Project Based Learning","score":0.5848000049591064,"subfield":{"id":"https://openalex.org/subfields/3304","display_name":"Education"},"field":{"id":"https://openalex.org/fields/33","display_name":"Social Sciences"},"domain":{"id":"https://openalex.org/domains/2","display_name":"Social Sciences"}},{"id":"https://openalex.org/T10906","display_name":"AI-based Problem Solving and Planning","score":0.5591999888420105,"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/modular-design","display_name":"Modular design","score":0.8401404619216919},{"id":"https://openalex.org/keywords/benchmark","display_name":"Benchmark (surveying)","score":0.8180521726608276},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.7342553734779358},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.4321502149105072},{"id":"https://openalex.org/keywords/machine-learning","display_name":"Machine learning","score":0.40097543597221375},{"id":"https://openalex.org/keywords/software-engineering","display_name":"Software engineering","score":0.35840243101119995},{"id":"https://openalex.org/keywords/programming-language","display_name":"Programming language","score":0.16977590322494507}],"concepts":[{"id":"https://openalex.org/C101468663","wikidata":"https://www.wikidata.org/wiki/Q1620158","display_name":"Modular design","level":2,"score":0.8401404619216919},{"id":"https://openalex.org/C185798385","wikidata":"https://www.wikidata.org/wiki/Q1161707","display_name":"Benchmark (surveying)","level":2,"score":0.8180521726608276},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7342553734779358},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4321502149105072},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.40097543597221375},{"id":"https://openalex.org/C115903868","wikidata":"https://www.wikidata.org/wiki/Q80993","display_name":"Software engineering","level":1,"score":0.35840243101119995},{"id":"https://openalex.org/C199360897","wikidata":"https://www.wikidata.org/wiki/Q9143","display_name":"Programming language","level":1,"score":0.16977590322494507},{"id":"https://openalex.org/C205649164","wikidata":"https://www.wikidata.org/wiki/Q1071","display_name":"Geography","level":0,"score":0.0},{"id":"https://openalex.org/C13280743","wikidata":"https://www.wikidata.org/wiki/Q131089","display_name":"Geodesy","level":1,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2025.acl-long.1499","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.acl-long.1499","pdf_url":"https://aclanthology.org/2025.acl-long.1499.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 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2025.acl-long.1499","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.acl-long.1499","pdf_url":"https://aclanthology.org/2025.acl-long.1499.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 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},"sustainable_development_goals":[{"display_name":"Sustainable cities and communities","score":0.6000000238418579,"id":"https://metadata.un.org/sdg/11"}],"awards":[{"id":"https://openalex.org/G6059964893","display_name":null,"funder_award_id":"STI 2030","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"},{"id":"https://openalex.org/G6812169155","display_name":null,"funder_award_id":"2023A1515140100","funder_id":"https://openalex.org/F4320337111","funder_display_name":"Basic and Applied Basic Research Foundation of Guangdong Province"},{"id":"https://openalex.org/G6948817555","display_name":null,"funder_award_id":"52308250","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"},{"id":"https://openalex.org/G8252633260","display_name":null,"funder_award_id":"2022ZD0208900","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"},{"id":"https://openalex.org/F4320337111","display_name":"Basic and Applied Basic Research Foundation of Guangdong Province","ror":null}],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4412889711.pdf","grobid_xml":"https://content.openalex.org/works/W4412889711.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W3088304681","https://openalex.org/W2961085424","https://openalex.org/W4306674287","https://openalex.org/W4387369504","https://openalex.org/W4394896187","https://openalex.org/W3170094116","https://openalex.org/W4386462264","https://openalex.org/W3107602296","https://openalex.org/W4364306694","https://openalex.org/W4312192474"],"abstract_inverted_index":{"One":[0],"of":[1,5,23,53,82,105,130],"the":[2,11,21,63,80,101,118,166],"research":[3],"focuses":[4],"large":[6],"language":[7],"models":[8,124],"(LLMs)":[9],"is":[10],"ability":[12,81],"to":[13,44,78,168,171],"generate":[14],"action":[15],"plans.Recent":[16],"studies":[17],"have":[18],"revealed":[19],"that":[20,58,116,138],"performance":[22],"LLMs":[24,83,106,140,170],"can":[25],"be":[26,60],"significantly":[27],"improved":[28],"by":[29],"integrating":[30],"external":[31],"tools.Based":[32],"on":[33],"this,":[34],"we":[35,98,163],"propose":[36],"a":[37,51,69,127],"benchmark":[38],"framework":[39,67],"called":[40],"PlanningArena,":[41,135],"which":[42],"aims":[43],"simulate":[45],"real":[46],"application":[47],"scenarios":[48],"and":[49,55,73,93,110,122,132,149,158],"provide":[50],"series":[52],"apps":[54],"API":[56],"tools":[57],"may":[59],"involved":[61],"in":[62,84,90,134,144],"actual":[64],"planning":[65,173],"process.This":[66],"adopts":[68],"modular":[70],"task":[71,102],"structure":[72],"combines":[74],"user":[75,95],"portrait":[76],"analysis":[77],"evaluate":[79],"correctly":[85],"selecting":[86],"tools,":[87],"logical":[88,145],"reasoning":[89],"complex":[91],"scenarios,":[92,157],"parsing":[94],"information.In":[96],"addition,":[97],"deeply":[99],"diagnose":[100],"execution":[103],"effect":[104],"from":[107],"both":[108],"macro":[109],"micro":[111],"levels.The":[112],"experimental":[113],"results":[114],"show":[115],"even":[117],"most":[119],"outstanding":[120],"GPT-4o":[121],"DeepSeekV3":[123],"only":[125],"achieved":[126],"total":[128],"score":[129],"56.5%":[131],"41.9%":[133],"respectively,":[136],"indicating":[137],"current":[139],"still":[141],"face":[142],"challenges":[143],"reasoning,":[146],"context":[147],"memory,":[148],"tool":[150],"calling":[151],"when":[152],"dealing":[153],"with":[154],"different":[155],"structures,":[156],"their":[159],"complexity.Through":[160],"this":[161],"benchmark,":[162],"further":[164],"explore":[165],"path":[167],"optimize":[169],"perform":[172],"tasks.":[174]},"counts_by_year":[{"year":2026,"cited_by_count":1}],"updated_date":"2026-06-11T09:08:48.828518","created_date":"2025-10-10T00:00:00"}