{"id":"https://openalex.org/W4416034758","doi":"https://doi.org/10.18653/v1/2025.findings-emnlp.776","title":"TRUEBench: Can LLM Response Meet Real-world Constraints as Productivity Assistant?","display_name":"TRUEBench: Can LLM Response Meet Real-world Constraints as Productivity Assistant?","publication_year":2025,"publication_date":"2025-01-01","ids":{"openalex":"https://openalex.org/W4416034758","doi":"https://doi.org/10.18653/v1/2025.findings-emnlp.776"},"language":null,"primary_location":{"id":"doi:10.18653/v1/2025.findings-emnlp.776","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-emnlp.776","pdf_url":"https://aclanthology.org/2025.findings-emnlp.776.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: EMNLP 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-emnlp.776.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5084928402","display_name":"J.\u2010H. Park","orcid":null},"institutions":[{"id":"https://openalex.org/I2250650973","display_name":"Samsung (South Korea)","ror":"https://ror.org/04w3jy968","country_code":"KR","type":"company","lineage":["https://openalex.org/I2250650973"]},{"id":"https://openalex.org/I2802194831","display_name":"Samsung Medical Center","ror":"https://ror.org/05a15z872","country_code":"KR","type":"healthcare","lineage":["https://openalex.org/I2250650973","https://openalex.org/I2802194831"]},{"id":"https://openalex.org/I4210088807","display_name":"Samsung Pharm (South Korea)","ror":"https://ror.org/003eds368","country_code":"KR","type":"company","lineage":["https://openalex.org/I4210088807"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Jiho Park","raw_affiliation_strings":["Samsung Research Seoul , South Korea"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Samsung Research Seoul , South Korea","institution_ids":["https://openalex.org/I2250650973","https://openalex.org/I2802194831","https://openalex.org/I4210088807"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Jongyoon Song","orcid":null},"institutions":[{"id":"https://openalex.org/I2250650973","display_name":"Samsung (South Korea)","ror":"https://ror.org/04w3jy968","country_code":"KR","type":"company","lineage":["https://openalex.org/I2250650973"]},{"id":"https://openalex.org/I2802194831","display_name":"Samsung Medical Center","ror":"https://ror.org/05a15z872","country_code":"KR","type":"healthcare","lineage":["https://openalex.org/I2250650973","https://openalex.org/I2802194831"]},{"id":"https://openalex.org/I4210088807","display_name":"Samsung Pharm (South Korea)","ror":"https://ror.org/003eds368","country_code":"KR","type":"company","lineage":["https://openalex.org/I4210088807"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Jongyoon Song","raw_affiliation_strings":["Samsung Research Seoul , South Korea"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Samsung Research Seoul , South Korea","institution_ids":["https://openalex.org/I2250650973","https://openalex.org/I2802194831","https://openalex.org/I4210088807"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5120056046","display_name":"Minjin Choi","orcid":null},"institutions":[{"id":"https://openalex.org/I2250650973","display_name":"Samsung (South Korea)","ror":"https://ror.org/04w3jy968","country_code":"KR","type":"company","lineage":["https://openalex.org/I2250650973"]},{"id":"https://openalex.org/I2802194831","display_name":"Samsung Medical Center","ror":"https://ror.org/05a15z872","country_code":"KR","type":"healthcare","lineage":["https://openalex.org/I2250650973","https://openalex.org/I2802194831"]},{"id":"https://openalex.org/I4210088807","display_name":"Samsung Pharm (South Korea)","ror":"https://ror.org/003eds368","country_code":"KR","type":"company","lineage":["https://openalex.org/I4210088807"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Minjin Choi","raw_affiliation_strings":["Samsung Research Seoul , South Korea"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Samsung Research Seoul , South Korea","institution_ids":["https://openalex.org/I2250650973","https://openalex.org/I2802194831","https://openalex.org/I4210088807"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5120056047","display_name":"Kyuho Heo","orcid":null},"institutions":[{"id":"https://openalex.org/I2250650973","display_name":"Samsung (South Korea)","ror":"https://ror.org/04w3jy968","country_code":"KR","type":"company","lineage":["https://openalex.org/I2250650973"]},{"id":"https://openalex.org/I2802194831","display_name":"Samsung Medical Center","ror":"https://ror.org/05a15z872","country_code":"KR","type":"healthcare","lineage":["https://openalex.org/I2250650973","https://openalex.org/I2802194831"]},{"id":"https://openalex.org/I4210088807","display_name":"Samsung Pharm (South Korea)","ror":"https://ror.org/003eds368","country_code":"KR","type":"company","lineage":["https://openalex.org/I4210088807"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Kyuho Heo","raw_affiliation_strings":["Samsung Research Seoul , South Korea"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Samsung Research Seoul , South Korea","institution_ids":["https://openalex.org/I2250650973","https://openalex.org/I2802194831","https://openalex.org/I4210088807"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5120056048","display_name":"Taehun Huh","orcid":null},"institutions":[{"id":"https://openalex.org/I2250650973","display_name":"Samsung (South Korea)","ror":"https://ror.org/04w3jy968","country_code":"KR","type":"company","lineage":["https://openalex.org/I2250650973"]},{"id":"https://openalex.org/I2802194831","display_name":"Samsung Medical Center","ror":"https://ror.org/05a15z872","country_code":"KR","type":"healthcare","lineage":["https://openalex.org/I2250650973","https://openalex.org/I2802194831"]},{"id":"https://openalex.org/I4210088807","display_name":"Samsung Pharm (South Korea)","ror":"https://ror.org/003eds368","country_code":"KR","type":"company","lineage":["https://openalex.org/I4210088807"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Taehun Huh","raw_affiliation_strings":["Samsung Research Seoul , South Korea"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Samsung Research Seoul , South Korea","institution_ids":["https://openalex.org/I2250650973","https://openalex.org/I2802194831","https://openalex.org/I4210088807"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5100754800","display_name":"Jungho Kim","orcid":"https://orcid.org/0000-0003-2241-4522"},"institutions":[{"id":"https://openalex.org/I2250650973","display_name":"Samsung (South Korea)","ror":"https://ror.org/04w3jy968","country_code":"KR","type":"company","lineage":["https://openalex.org/I2250650973"]},{"id":"https://openalex.org/I2802194831","display_name":"Samsung Medical Center","ror":"https://ror.org/05a15z872","country_code":"KR","type":"healthcare","lineage":["https://openalex.org/I2250650973","https://openalex.org/I2802194831"]},{"id":"https://openalex.org/I4210088807","display_name":"Samsung Pharm (South Korea)","ror":"https://ror.org/003eds368","country_code":"KR","type":"company","lineage":["https://openalex.org/I4210088807"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Ji Won Kim","raw_affiliation_strings":["Samsung Research Seoul , South Korea"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Samsung Research Seoul , South Korea","institution_ids":["https://openalex.org/I2250650973","https://openalex.org/I2802194831","https://openalex.org/I4210088807"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":3,"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":0.37210332,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"14383","last_page":"14409"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T12127","display_name":"Software System Performance and Reliability","score":0.11779999732971191,"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/T12127","display_name":"Software System Performance and Reliability","score":0.11779999732971191,"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/T11596","display_name":"Constraint Satisfaction and Optimization","score":0.054999999701976776,"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/T13197","display_name":"Spreadsheets and End-User Computing","score":0.04560000076889992,"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"}}],"keywords":[{"id":"https://openalex.org/keywords/productivity","display_name":"Productivity","score":0.6373000144958496},{"id":"https://openalex.org/keywords/production","display_name":"Production (economics)","score":0.412200003862381},{"id":"https://openalex.org/keywords/quality","display_name":"Quality (philosophy)","score":0.2962000072002411},{"id":"https://openalex.org/keywords/measure","display_name":"Measure (data warehouse)","score":0.2757999897003174},{"id":"https://openalex.org/keywords/yield","display_name":"Yield (engineering)","score":0.2705000042915344}],"concepts":[{"id":"https://openalex.org/C204983608","wikidata":"https://www.wikidata.org/wiki/Q2111958","display_name":"Productivity","level":2,"score":0.6373000144958496},{"id":"https://openalex.org/C2778348673","wikidata":"https://www.wikidata.org/wiki/Q739302","display_name":"Production (economics)","level":2,"score":0.412200003862381},{"id":"https://openalex.org/C162324750","wikidata":"https://www.wikidata.org/wiki/Q8134","display_name":"Economics","level":0,"score":0.35260000824928284},{"id":"https://openalex.org/C2779530757","wikidata":"https://www.wikidata.org/wiki/Q1207505","display_name":"Quality (philosophy)","level":2,"score":0.2962000072002411},{"id":"https://openalex.org/C2780009758","wikidata":"https://www.wikidata.org/wiki/Q6804172","display_name":"Measure (data warehouse)","level":2,"score":0.2757999897003174},{"id":"https://openalex.org/C134121241","wikidata":"https://www.wikidata.org/wiki/Q899301","display_name":"Yield (engineering)","level":2,"score":0.2705000042915344},{"id":"https://openalex.org/C144133560","wikidata":"https://www.wikidata.org/wiki/Q4830453","display_name":"Business","level":0,"score":0.26510000228881836},{"id":"https://openalex.org/C18762648","wikidata":"https://www.wikidata.org/wiki/Q42213","display_name":"Work (physics)","level":2,"score":0.2635999917984009},{"id":"https://openalex.org/C2776036281","wikidata":"https://www.wikidata.org/wiki/Q48769818","display_name":"Constraint (computer-aided design)","level":2,"score":0.2619999945163727},{"id":"https://openalex.org/C175605778","wikidata":"https://www.wikidata.org/wiki/Q3299701","display_name":"Natural resource economics","level":1,"score":0.25839999318122864},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.2533000111579895}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2025.findings-emnlp.776","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-emnlp.776","pdf_url":"https://aclanthology.org/2025.findings-emnlp.776.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: EMNLP 2025","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2025.findings-emnlp.776","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-emnlp.776","pdf_url":"https://aclanthology.org/2025.findings-emnlp.776.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: EMNLP 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/W4416034758.pdf","grobid_xml":"https://content.openalex.org/works/W4416034758.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Large":[0],"language":[1],"models":[2],"(LLMs)":[3],"are":[4],"increasingly":[5],"integral":[6],"as":[7],"productivity":[8,74,160],"assistants,":[9],"but":[10],"existing":[11,134],"benchmarks":[12,22],"fall":[13],"short":[14],"in":[15,36,116,158],"rigorously":[16],"evaluating":[17],"their":[18,163],"real-world":[19],"instruction-following":[20],"capabilities.Current":[21],"often":[23],"(i)":[24],"lack":[25],"sufficient":[26],"multilinguality,":[27],"(ii)":[28],"fail":[29],"to":[30,93,113],"capture":[31,94],"the":[32,42],"implicit":[33,98],"constraints":[34,109,120],"inherent":[35],"user":[37],"requests,":[38],"and":[39,51,97,100,110,153,165],"(iii)":[40],"overlook":[41],"complexities":[43],"of":[44,156],"multiturn":[45,103],"dialogue.To":[46],"address":[47],"these":[48],"critical":[49],"gaps":[50],"provide":[52],"a":[53,67,138,146,151],"more":[54],"realistic":[55,154],"assessment,":[56],"we":[57,118],"introduce":[58],"TRUEBench":[59,128],"(Trustworthy":[60],"Real-world":[61],"Usage":[62],"Evaluation":[63],"Benchmark)":[64],"1":[65],",":[66],"novel":[68],"benchmark":[69],"specifically":[70],"designed":[71],"for":[72,136],"LLM-based":[73],"assistants.TRUEBench":[75],"distinguishes":[76],"itself":[77],"by":[78],"featuring":[79],"input":[80],"prompts":[81],"across":[82],"12":[83],"languages,":[84],"incorporating":[85],"intra-instance":[86],"multilingual":[87],"instructions,":[88],"employing":[89],"rigorous":[90],"evaluation":[91],"criteria":[92],"both":[95,107],"explicit":[96],"constraints,":[99],"including":[101],"complex":[102],"dialogue":[104],"scenarios":[105],"with":[106],"accumulating":[108],"context":[111],"switches.Furthermore,":[112],"ensure":[114],"reliability":[115],"evaluation,":[117],"refined":[119],"using":[121],"an":[122],"LLM":[123],"validator.Extensive":[124],"experiments":[125],"demonstrate":[126],"that":[127],"presents":[129],"significantly":[130],"greater":[131],"challenges":[132],"than":[133],"benchmarks;":[135],"instance,":[137],"strong":[139],"model":[140],"like":[141],"OpenAI":[142],"o1":[143],"achieved":[144],"only":[145],"69.07%overall":[147],"pass":[148],"rate.TRUEBench":[149],"offers":[150],"demanding":[152],"assessment":[155],"LLMs":[157],"practical":[159],"settings,":[161],"highlighting":[162],"capabilities":[164],"limitations.":[166]},"counts_by_year":[],"updated_date":"2026-08-23T07:36:19.812096","created_date":"2025-11-08T00:00:00"}