{"id":"https://openalex.org/W4412889602","doi":"https://doi.org/10.18653/v1/2025.acl-long.1576","title":"EducationQ: Evaluating LLMs\u2019 Teaching Capabilities Through Multi-Agent Dialogue Framework","display_name":"EducationQ: Evaluating LLMs\u2019 Teaching Capabilities Through Multi-Agent Dialogue Framework","publication_year":2025,"publication_date":"2025-01-01","ids":{"openalex":"https://openalex.org/W4412889602","doi":"https://doi.org/10.18653/v1/2025.acl-long.1576"},"language":"en","primary_location":{"id":"doi:10.18653/v1/2025.acl-long.1576","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.acl-long.1576","pdf_url":"https://aclanthology.org/2025.acl-long.1576.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":"conference-paper","indexed_in":["arxiv","crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2025.acl-long.1576.pdf","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5112127173","display_name":"Yao Shi","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yao Shi","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5119181812","display_name":"Rongkeng Liang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Rongkeng Liang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5100563672","display_name":"Yong Xu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yong Xu","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":null,"has_fulltext":true,"cited_by_count":4,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"32799","last_page":"32828"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10456","display_name":"Multi-Agent Systems and Negotiation","score":0.8528000116348267,"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/T10456","display_name":"Multi-Agent Systems and Negotiation","score":0.8528000116348267,"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/T13643","display_name":"Artificial Intelligence in Law","score":0.7620999813079834,"subfield":{"id":"https://openalex.org/subfields/3320","display_name":"Political Science and International Relations"},"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/T10215","display_name":"Semantic Web and Ontologies","score":0.7254999876022339,"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/computer-science","display_name":"Computer science","score":0.5799376368522644}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5799376368522644}],"mesh":[],"locations_count":2,"locations":[{"id":"doi:10.18653/v1/2025.acl-long.1576","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.acl-long.1576","pdf_url":"https://aclanthology.org/2025.acl-long.1576.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"},{"id":"pmh:oai:arXiv.org:2504.14928","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2504.14928","pdf_url":"https://arxiv.org/pdf/2504.14928","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"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":"cc-by-sa","license_id":"https://openalex.org/licenses/cc-by-sa","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"}],"best_oa_location":{"id":"doi:10.18653/v1/2025.acl-long.1576","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.acl-long.1576","pdf_url":"https://aclanthology.org/2025.acl-long.1576.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":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4412889602.pdf"},"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":{"Large":[0],"language":[1],"models":[2,96,140],"(LLMs)":[3],"increasingly":[4],"serve":[5],"as":[6],"educational":[7,43,178],"tools,":[8],"yet":[9],"evaluating":[10],"their":[11],"teaching":[12,38,78,102,161],"capabilities":[13,39,90],"remains":[14],"challenging":[15],"due":[16],"to":[17],"the":[18],"resource-intensive,":[19],"context-dependent,":[20],"and":[21,51,64,72,129],"methodologically":[22],"complex":[23],"nature":[24],"of":[25,159,183],"teacher-student":[26],"interactions.":[27],"We":[28],"introduce":[29],"EducationQ,":[30],"a":[31,107],"multi-agent":[32],"dialogue":[33],"framework":[34],"that":[35,77,113,168],"efficiently":[36],"assesses":[37],"through":[40],"simulated":[41],"dynamic":[42],"scenarios,":[44],"featuring":[45],"specialized":[46,171],"agents":[47],"for":[48],"teaching,":[49],"learning,":[50],"evaluation.":[52],"Testing":[53],"14":[54],"LLMs":[55],"across":[56],"major":[57],"AI":[58,179],"Organizations":[59],"(OpenAI,":[60],"Meta,":[61],"Google,":[62],"Anthropic,":[63],"others)":[65],"on":[66],"1,498":[67],"questions":[68],"spanning":[69],"13":[70],"disciplines":[71],"10":[73],"difficulty":[74],"levels":[75],"reveals":[76],"effectiveness":[79],"does":[80],"not":[81],"correlate":[82],"linearly":[83],"with":[84,92,126,154],"model":[85],"scale":[86],"or":[87],"general":[88],"reasoning":[89],"-":[91],"some":[93],"smaller":[94],"open-source":[95],"outperforming":[97],"larger":[98],"commercial":[99],"counterparts":[100],"in":[101,110],"contexts.":[103],"This":[104],"finding":[105],"highlights":[106],"critical":[108],"gap":[109],"current":[111],"evaluations":[112,150],"prioritize":[114,180],"knowledge":[115],"recall":[116],"over":[117],"interactive":[118],"pedagogy.":[119],"Our":[120],"mixed-methods":[121],"evaluation,":[122],"combining":[123],"quantitative":[124],"metrics":[125],"qualitative":[127,157],"analysis":[128,158],"expert":[130,149],"case":[131],"studies,":[132],"identifies":[133],"distinct":[134],"pedagogical":[135,185],"strengths":[136],"employed":[137],"by":[138],"top-performing":[139],"(e.g.,":[141],"sophisticated":[142],"questioning":[143],"strategies,":[144],"adaptive":[145],"feedback":[146],"mechanisms).":[147],"Human":[148],"show":[151],"78%":[152],"agreement":[153],"our":[155,164],"automated":[156],"effective":[160],"behaviors,":[162],"validating":[163],"methodology.":[165],"EducationQ":[166],"demonstrates":[167],"LLMs-as-teachers":[169],"require":[170],"optimization":[172],"beyond":[173],"simple":[174],"scaling,":[175],"suggesting":[176],"next-generation":[177],"targeted":[181],"enhancement":[182],"specific":[184],"effectiveness.":[186]},"counts_by_year":[{"year":2026,"cited_by_count":4}],"updated_date":"2026-07-14T23:27:15.235271","created_date":"2025-10-10T00:00:00"}