{"id":"https://openalex.org/W4406461795","doi":"https://doi.org/10.1109/slt61566.2024.10832237","title":"Confidence Estimation For LLM-Based Dialogue State Tracking","display_name":"Confidence Estimation For LLM-Based Dialogue State Tracking","publication_year":2024,"publication_date":"2024-12-02","ids":{"openalex":"https://openalex.org/W4406461795","doi":"https://doi.org/10.1109/slt61566.2024.10832237"},"language":"en","primary_location":{"id":"doi:10.1109/slt61566.2024.10832237","is_oa":false,"landing_page_url":"https://doi.org/10.1109/slt61566.2024.10832237","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/A5108133777","display_name":"Y. Sun","orcid":"https://orcid.org/0009-0001-7929-9003"},"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":"Yi-Jyun Sun","raw_affiliation_strings":["University of Illinois Urbana-Champaign"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Illinois Urbana-Champaign","institution_ids":["https://openalex.org/I157725225"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5007964652","display_name":"Suvodip Dey","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":"Suvodip Dey","raw_affiliation_strings":["University of Illinois Urbana-Champaign"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Illinois 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 Urbana-Champaign"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Illinois Urbana-Champaign","institution_ids":["https://openalex.org/I157725225"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5087941479","display_name":"G\u00f6khan T\u00fcr","orcid":"https://orcid.org/0009-0008-7740-2557"},"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 Urbana-Champaign"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Illinois Urbana-Champaign","institution_ids":["https://openalex.org/I157725225"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I157725225"],"apc_list":null,"apc_paid":null,"fwci":1.1943,"has_fulltext":false,"cited_by_count":4,"citation_normalized_percentile":{"value":0.8372839,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":95,"max":97},"biblio":{"volume":null,"issue":null,"first_page":"1083","last_page":"1090"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T12031","display_name":"Speech and dialogue systems","score":0.9994000196456909,"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":0.9994000196456909,"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":0.9952999949455261,"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":0.9750000238418579,"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":0.6414535045623779},{"id":"https://openalex.org/keywords/estimation","display_name":"Estimation","score":0.5622316598892212},{"id":"https://openalex.org/keywords/state","display_name":"State (computer science)","score":0.5339934229850769},{"id":"https://openalex.org/keywords/tracking","display_name":"Tracking (education)","score":0.5165178179740906},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.4530491530895233},{"id":"https://openalex.org/keywords/psychology","display_name":"Psychology","score":0.13001862168312073},{"id":"https://openalex.org/keywords/engineering","display_name":"Engineering","score":0.12861040234565735},{"id":"https://openalex.org/keywords/algorithm","display_name":"Algorithm","score":0.10154199600219727}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6414535045623779},{"id":"https://openalex.org/C96250715","wikidata":"https://www.wikidata.org/wiki/Q965330","display_name":"Estimation","level":2,"score":0.5622316598892212},{"id":"https://openalex.org/C48103436","wikidata":"https://www.wikidata.org/wiki/Q599031","display_name":"State (computer science)","level":2,"score":0.5339934229850769},{"id":"https://openalex.org/C2775936607","wikidata":"https://www.wikidata.org/wiki/Q466845","display_name":"Tracking (education)","level":2,"score":0.5165178179740906},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4530491530895233},{"id":"https://openalex.org/C15744967","wikidata":"https://www.wikidata.org/wiki/Q9418","display_name":"Psychology","level":0,"score":0.13001862168312073},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"score":0.12861040234565735},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.10154199600219727},{"id":"https://openalex.org/C201995342","wikidata":"https://www.wikidata.org/wiki/Q682496","display_name":"Systems engineering","level":1,"score":0.0},{"id":"https://openalex.org/C19417346","wikidata":"https://www.wikidata.org/wiki/Q7922","display_name":"Pedagogy","level":1,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/slt61566.2024.10832237","is_oa":false,"landing_page_url":"https://doi.org/10.1109/slt61566.2024.10832237","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/F4320332815","display_name":"Advanced Research Projects Agency","ror":"https://ror.org/02caytj08"}],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":32,"referenced_works":["https://openalex.org/W2964006684","https://openalex.org/W3034284249","https://openalex.org/W3045703328","https://openalex.org/W3100128199","https://openalex.org/W3168491067","https://openalex.org/W3196636639","https://openalex.org/W4206496297","https://openalex.org/W4281557260","https://openalex.org/W4285429195","https://openalex.org/W4290994954","https://openalex.org/W4321594373","https://openalex.org/W4322718191","https://openalex.org/W4379928343","https://openalex.org/W4383468784","https://openalex.org/W4385570088","https://openalex.org/W4385573072","https://openalex.org/W4387687622","https://openalex.org/W4389010465","https://openalex.org/W4389518686","https://openalex.org/W4389520749","https://openalex.org/W4402670114","https://openalex.org/W6791446462","https://openalex.org/W6796581206","https://openalex.org/W6838865847","https://openalex.org/W6839548382","https://openalex.org/W6844238458","https://openalex.org/W6850423384","https://openalex.org/W6850625674","https://openalex.org/W6853260906","https://openalex.org/W6853359188","https://openalex.org/W6853562202","https://openalex.org/W6853955615"],"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":{"Estimation":[0],"of":[1,35,56,72,107,152],"a":[2,105,132],"model\u2019s":[3],"confidence":[4,77,94,176],"on":[5,15,61,97],"its":[6],"outputs":[7],"is":[8],"critical":[9],"for":[10,21,40,92,136,149],"Conversational":[11],"AI":[12],"systems":[13,69],"based":[14,96],"large":[16],"language":[17],"models":[18],"(LLMs),":[19],"especially":[20],"reducing":[22],"hallucination":[23],"and":[24,42,48,104],"preventing":[25],"over-reliance.":[26],"In":[27],"this":[28],"work,":[29],"we":[30,140],"provide":[31],"an":[32,145],"exhaustive":[33],"exploration":[34],"methods,":[36,109],"including":[37],"approaches":[38],"proposed":[39,135],"open-":[41],"closed-weight":[43],"LLMs,":[44],"aimed":[45],"at":[46],"quantifying":[47],"leveraging":[49],"model":[50,74,86,147],"uncertainty":[51],"to":[52,81,118],"improve":[53],"the":[54,73,111,114,150],"reliability":[55],"LLM-generated":[57],"responses,":[58],"specifically":[59],"focusing":[60],"dialogue":[62,68],"state":[63],"tracking":[64],"(DST)":[65],"in":[66,170],"task-oriented":[67],"(TODS).":[70],"Regardless":[71],"type,":[75],"well-calibrated":[76],"scores":[78,95],"are":[79],"essential":[80],"handle":[82],"uncertainties,":[83],"thereby":[84],"improving":[85],"performance.":[87],"We":[88,127],"evaluate":[89],"four":[90],"methods":[91,143],"estimating":[93],"softmax,":[98],"raw":[99],"token":[100],"scores,":[101],"verbalized":[102],"confidences,":[103],"combination":[106],"these":[108,130,142],"using":[110,144],"area":[112],"under":[113],"curve":[115],"(AUC)":[116],"metric":[117],"assess":[119,141],"calibration,":[120],"with":[121,131],"higher":[122],"AUC":[123,172],"indicating":[124,174],"better":[125,175],"calibration.":[126,178],"also":[128,162],"enhance":[129],"self-probing":[133],"mechanism,":[134],"closed":[137],"models.":[138],"Furthermore,":[139],"open-weight":[146,166],"fine-tuned":[148],"task":[151],"DST,":[153],"achieving":[154],"superior":[155],"joint":[156],"goal":[157],"accuracy":[158],"(JGA).":[159],"Our":[160],"findings":[161],"suggest":[163],"that":[164],"fine-tuning":[165],"LLMs":[167],"can":[168],"result":[169],"enhanced":[171],"performance,":[173],"score":[177]},"counts_by_year":[{"year":2026,"cited_by_count":1},{"year":2025,"cited_by_count":3}],"updated_date":"2026-06-26T08:34:08.712188","created_date":"2025-10-10T00:00:00"}