{"id":"https://openalex.org/W4406461864","doi":"https://doi.org/10.1109/slt61566.2024.10832298","title":"Large Language Models as User-Agents For Evaluating Task-Oriented-Dialogue Systems","display_name":"Large Language Models as User-Agents For Evaluating Task-Oriented-Dialogue Systems","publication_year":2024,"publication_date":"2024-12-02","ids":{"openalex":"https://openalex.org/W4406461864","doi":"https://doi.org/10.1109/slt61566.2024.10832298"},"language":"en","primary_location":{"id":"doi:10.1109/slt61566.2024.10832298","is_oa":false,"landing_page_url":"https://doi.org/10.1109/slt61566.2024.10832298","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2024 IEEE Spoken Language Technology Workshop (SLT)","raw_type":"proceedings-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":false,"oa_status":"closed","oa_url":null,"any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5019361560","display_name":"Taaha Kazi","orcid":null},"institutions":[{"id":"https://openalex.org/I157725225","display_name":"University of Illinois Urbana-Champaign","ror":"https://ror.org/047426m28","country_code":"US","type":"education","lineage":["https://openalex.org/I157725225"]}],"countries":["US"],"is_corresponding":true,"raw_author_name":"Taaha Kazi","raw_affiliation_strings":["University of Illinois at Urbana Champaign"],"affiliations":[{"raw_affiliation_string":"University of Illinois at Urbana Champaign","institution_ids":["https://openalex.org/I157725225"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5075359226","display_name":"Ruiliang Lyu","orcid":null},"institutions":[{"id":"https://openalex.org/I157725225","display_name":"University of Illinois Urbana-Champaign","ror":"https://ror.org/047426m28","country_code":"US","type":"education","lineage":["https://openalex.org/I157725225"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Ruiliang Lyu","raw_affiliation_strings":["University of Illinois at Urbana Champaign"],"affiliations":[{"raw_affiliation_string":"University of Illinois at Urbana Champaign","institution_ids":["https://openalex.org/I157725225"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5040682019","display_name":"Sizhe Zhou","orcid":"https://orcid.org/0009-0005-5145-7152"},"institutions":[{"id":"https://openalex.org/I157725225","display_name":"University of Illinois Urbana-Champaign","ror":"https://ror.org/047426m28","country_code":"US","type":"education","lineage":["https://openalex.org/I157725225"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Sizhe Zhou","raw_affiliation_strings":["University of Illinois at Urbana Champaign"],"affiliations":[{"raw_affiliation_string":"University of Illinois at Urbana Champaign","institution_ids":["https://openalex.org/I157725225"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5068709817","display_name":"Dilek Hakkani\u2010T\u00fcr","orcid":null},"institutions":[{"id":"https://openalex.org/I157725225","display_name":"University of Illinois Urbana-Champaign","ror":"https://ror.org/047426m28","country_code":"US","type":"education","lineage":["https://openalex.org/I157725225"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Dilek Hakkani-T\u00fcr","raw_affiliation_strings":["University of Illinois at Urbana Champaign"],"affiliations":[{"raw_affiliation_string":"University of Illinois at Urbana Champaign","institution_ids":["https://openalex.org/I157725225"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5087941479","display_name":"G\u00f6khan T\u00fcr","orcid":null},"institutions":[{"id":"https://openalex.org/I157725225","display_name":"University of Illinois Urbana-Champaign","ror":"https://ror.org/047426m28","country_code":"US","type":"education","lineage":["https://openalex.org/I157725225"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Gokhan Tur","raw_affiliation_strings":["University of Illinois at Urbana Champaign"],"affiliations":[{"raw_affiliation_string":"University of Illinois at Urbana Champaign","institution_ids":["https://openalex.org/I157725225"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":5,"corresponding_author_ids":["https://openalex.org/A5019361560"],"corresponding_institution_ids":["https://openalex.org/I157725225"],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":{"value":0.23819287,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"913","last_page":"920"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T12031","display_name":"Speech and dialogue systems","score":0.9998000264167786,"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":0.9998000264167786,"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":0.9969000220298767,"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/T12128","display_name":"AI in Service Interactions","score":0.991599977016449,"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":0.8199211955070496},{"id":"https://openalex.org/keywords/task","display_name":"Task (project management)","score":0.633184552192688},{"id":"https://openalex.org/keywords/human\u2013computer-interaction","display_name":"Human\u2013computer interaction","score":0.5721525549888611},{"id":"https://openalex.org/keywords/programming-language","display_name":"Programming language","score":0.3503347635269165},{"id":"https://openalex.org/keywords/systems-engineering","display_name":"Systems engineering","score":0.18035182356834412},{"id":"https://openalex.org/keywords/engineering","display_name":"Engineering","score":0.08386766910552979}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8199211955070496},{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.633184552192688},{"id":"https://openalex.org/C107457646","wikidata":"https://www.wikidata.org/wiki/Q207434","display_name":"Human\u2013computer interaction","level":1,"score":0.5721525549888611},{"id":"https://openalex.org/C199360897","wikidata":"https://www.wikidata.org/wiki/Q9143","display_name":"Programming language","level":1,"score":0.3503347635269165},{"id":"https://openalex.org/C201995342","wikidata":"https://www.wikidata.org/wiki/Q682496","display_name":"Systems engineering","level":1,"score":0.18035182356834412},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"score":0.08386766910552979}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/slt61566.2024.10832298","is_oa":false,"landing_page_url":"https://doi.org/10.1109/slt61566.2024.10832298","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2024 IEEE Spoken Language Technology Workshop (SLT)","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[],"awards":[],"funders":[{"id":"https://openalex.org/F4320307764","display_name":"Microsoft","ror":"https://ror.org/00d0nc645"},{"id":"https://openalex.org/F4320332180","display_name":"Defense Advanced Research Projects Agency","ror":"https://ror.org/02caytj08"}],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":42,"referenced_works":["https://openalex.org/W2055766585","https://openalex.org/W2109038907","https://openalex.org/W2150407103","https://openalex.org/W2799002257","https://openalex.org/W2891732163","https://openalex.org/W2896457183","https://openalex.org/W2963567240","https://openalex.org/W2963797754","https://openalex.org/W3088273075","https://openalex.org/W3207280389","https://openalex.org/W4205480693","https://openalex.org/W4288089799","https://openalex.org/W4322718191","https://openalex.org/W4365601361","https://openalex.org/W4377238805","https://openalex.org/W4378189609","https://openalex.org/W4384918448","https://openalex.org/W4385573592","https://openalex.org/W4385682136","https://openalex.org/W4387074860","https://openalex.org/W4389009464","https://openalex.org/W4389010469","https://openalex.org/W4389010534","https://openalex.org/W4391591592","https://openalex.org/W4402683861","https://openalex.org/W4404783034","https://openalex.org/W6755207826","https://openalex.org/W6769627184","https://openalex.org/W6771489201","https://openalex.org/W6778883912","https://openalex.org/W6782879696","https://openalex.org/W6802421032","https://openalex.org/W6809646742","https://openalex.org/W6810640255","https://openalex.org/W6838865847","https://openalex.org/W6849590751","https://openalex.org/W6850625674","https://openalex.org/W6852874933","https://openalex.org/W6854866820","https://openalex.org/W6856086692","https://openalex.org/W6856919264","https://openalex.org/W6861272002"],"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/W3196817267","https://openalex.org/W1976600725"],"abstract_inverted_index":{"Traditionally,":[0],"offline":[1],"datasets":[2,13],"have":[3],"been":[4],"used":[5],"to":[6,53,64],"evaluate":[7],"task-oriented":[8],"dialogue":[9],"(TOD)":[10],"models.":[11],"These":[12],"lack":[14],"context":[15],"awareness,":[16],"making":[17,39],"them":[18,40],"suboptimal":[19],"benchmarks":[20],"for":[21,67,96,112],"conversational":[22],"systems.":[23,72],"In":[24],"contrast,":[25],"user-agents,":[26],"which":[27],"are":[28],"context-aware,":[29],"can":[30],"simulate":[31],"the":[32,68,85,97,103,113],"variability":[33],"and":[34,83,92],"unpredictability":[35],"of":[36,70,90,105,116],"human":[37],"conversations,":[38],"better":[41,106],"alternatives":[42],"as":[43,81],"evaluators.":[44],"Prior":[45],"research":[46],"has":[47],"utilized":[48],"large":[49],"language":[50],"models":[51,118],"(LLMs)":[52],"develop":[54],"user-agents.":[55],"Our":[56,88],"work":[57],"builds":[58],"upon":[59],"this":[60,120],"by":[61],"using":[62,78],"LLMs":[63],"create":[65],"user-agents":[66,98],"evaluation":[69,89,115],"TOD":[71,117],"This":[73],"involves":[74],"prompting":[75],"an":[76],"LLM,":[77],"in-context":[79],"examples":[80],"guidance,":[82],"tracking":[84],"user-goal":[86],"state.":[87],"diversity":[91],"task":[93],"completion":[94],"metrics":[95],"shows":[99],"improved":[100],"performance":[101],"with":[102],"use":[104],"prompts.":[107],"Additionally,":[108],"we":[109],"propose":[110],"methodologies":[111],"automatic":[114],"within":[119],"dynamic":[121],"framework.":[122],"We":[123],"make":[124],"our":[125],"code":[126],"publicly":[127],"available<sup":[128],"xmlns:mml=\"http://www.w3.org/1998/Math/MathML\"":[129,131],"xmlns:xlink=\"http://www.w3.org/1999/xlink\">1</sup><sup":[130],"xmlns:xlink=\"http://www.w3.org/1999/xlink\">1</sup>https://github.com/TaahaKazi/user-agent":[132]},"counts_by_year":[],"updated_date":"2025-11-06T03:46:38.306776","created_date":"2025-10-10T00:00:00"}