{"id":"https://openalex.org/W4412888099","doi":"https://doi.org/10.18653/v1/2025.findings-acl.860","title":"CHARPEVAL: Benchmarking Large Language Models\u2019 Contextual Reasoning in Knowledge-Grounded Dialogue","display_name":"CHARPEVAL: Benchmarking Large Language Models\u2019 Contextual Reasoning in Knowledge-Grounded Dialogue","publication_year":2025,"publication_date":"2025-01-01","ids":{"openalex":"https://openalex.org/W4412888099","doi":"https://doi.org/10.18653/v1/2025.findings-acl.860"},"language":"en","primary_location":{"id":"doi:10.18653/v1/2025.findings-acl.860","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-acl.860","pdf_url":"https://aclanthology.org/2025.findings-acl.860.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 2025","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2025.findings-acl.860.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5059337634","display_name":"Abbas Ghaddar","orcid":"https://orcid.org/0000-0001-5446-2479"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Abbas Ghaddar","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5064431976","display_name":"David Alfonso-Hermelo","orcid":"https://orcid.org/0009-0009-4591-3077"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"David Alfonso-Hermelo","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5034154991","display_name":"Philippe Langlais","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Philippe Langlais","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5019715118","display_name":"Boxing Chen","orcid":"https://orcid.org/0000-0002-3170-4858"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Boxing Chen","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5050015287","display_name":"Prasanna Parthasarathi","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Prasanna Parthasarathi","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":0,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"16764","last_page":"16775"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T12031","display_name":"Speech and dialogue systems","score":0.9783999919891357,"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/T12031","display_name":"Speech and dialogue systems","score":0.9783999919891357,"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/T10028","display_name":"Topic Modeling","score":0.9664000272750854,"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/T10456","display_name":"Multi-Agent Systems and Negotiation","score":0.9552000164985657,"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":0.7358835935592651},{"id":"https://openalex.org/keywords/benchmarking","display_name":"Benchmarking","score":0.7351586222648621},{"id":"https://openalex.org/keywords/natural-language-processing","display_name":"Natural language processing","score":0.548901379108429},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.4601818323135376}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7358835935592651},{"id":"https://openalex.org/C86251818","wikidata":"https://www.wikidata.org/wiki/Q816754","display_name":"Benchmarking","level":2,"score":0.7351586222648621},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.548901379108429},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4601818323135376},{"id":"https://openalex.org/C162853370","wikidata":"https://www.wikidata.org/wiki/Q39809","display_name":"Marketing","level":1,"score":0.0},{"id":"https://openalex.org/C144133560","wikidata":"https://www.wikidata.org/wiki/Q4830453","display_name":"Business","level":0,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2025.findings-acl.860","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-acl.860","pdf_url":"https://aclanthology.org/2025.findings-acl.860.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 2025","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2025.findings-acl.860","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-acl.860","pdf_url":"https://aclanthology.org/2025.findings-acl.860.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 2025","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4412888099.pdf","grobid_xml":"https://content.openalex.org/works/W4412888099.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/W4238897586","https://openalex.org/W435179959","https://openalex.org/W2619091065","https://openalex.org/W2059640416","https://openalex.org/W1490753184","https://openalex.org/W2284465472","https://openalex.org/W3204019825"],"abstract_inverted_index":{"This":[0],"paper":[1],"presents":[2],"CHARPEVAL,":[3],"a":[4,47],"challenging":[5],"benchmark":[6],"specifically":[7],"designed":[8],"to":[9,18,60,63,87],"evaluate":[10],"the":[11,29,41,72,85],"ability":[12],"of":[13,50,69],"Large":[14],"Language":[15],"Models":[16],"(LLMs)":[17],"perform":[19],"contextualized":[20],"reasoning":[21],"in":[22],"knowledge-grounded":[23],"dialogue":[24],"scenarios.The":[25],"task":[26],"involves":[27],"selecting":[28],"correct":[30],"response":[31],"from":[32],"6":[33],"options,":[34],"including":[35],"5":[36],"manually":[37],"crafted":[38],"distractors,":[39],"given":[40],"conversation":[42],"history.Extensive":[43],"benchmarking":[44],"experiments":[45],"with":[46,81],"diverse":[48],"set":[49],"state-of-the-art":[51],"open-weight":[52],"LLMs":[53,89],"show":[54],"poor":[55],"performance":[56],"on":[57],"CHARPEVAL":[58],"due":[59],"their":[61],"inability":[62],"effectively":[64],"reason":[65],"over":[66],"discontinuous":[67],"chunks":[68],"text":[70],"across":[71,79],"input.Our":[73],"analysis":[74],"reveals":[75],"systematic":[76],"error":[77],"patterns":[78],"models":[80],"different":[82],"properties,":[83],"highlighting":[84],"need":[86],"improve":[88],"beyond":[90],"simply":[91],"scaling-up":[92],"data":[93],"and":[94],"compute.CHARPEVAL":[95],"is":[96],"publicly":[97],"available":[98],"at":[99]},"counts_by_year":[],"updated_date":"2026-07-14T23:27:15.235271","created_date":"2025-10-10T00:00:00"}