{"id":"https://openalex.org/W4412887744","doi":"https://doi.org/10.18653/v1/2025.findings-acl.1049","title":"Localizing and Mitigating Errors in Long-form Question Answering","display_name":"Localizing and Mitigating Errors in Long-form Question Answering","publication_year":2025,"publication_date":"2025-01-01","ids":{"openalex":"https://openalex.org/W4412887744","doi":"https://doi.org/10.18653/v1/2025.findings-acl.1049"},"language":"en","primary_location":{"id":"doi:10.18653/v1/2025.findings-acl.1049","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-acl.1049","pdf_url":"https://aclanthology.org/2025.findings-acl.1049.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.1049.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5065230808","display_name":"Rachneet Singh Sachdeva","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Rachneet Singh Sachdeva","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5102231254","display_name":"Yixiao Song","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yixiao Song","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5082767919","display_name":"Mohit Iyyer","orcid":"https://orcid.org/0000-0001-7340-0804"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Mohit Iyyer","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5027450194","display_name":"Iryna Gurevych","orcid":"https://orcid.org/0000-0003-2187-7621"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Iryna Gurevych","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":0.0,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":{"value":0.14199671,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"20437","last_page":"20469"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.9850000143051147,"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/T10028","display_name":"Topic Modeling","score":0.9850000143051147,"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/T13018","display_name":"Seismology and Earthquake Studies","score":0.9713000059127808,"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/T12031","display_name":"Speech and dialogue systems","score":0.9649999737739563,"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/question-answering","display_name":"Question answering","score":0.6964258551597595},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.590765118598938},{"id":"https://openalex.org/keywords/information-retrieval","display_name":"Information retrieval","score":0.3787664771080017},{"id":"https://openalex.org/keywords/natural-language-processing","display_name":"Natural language processing","score":0.3754114806652069}],"concepts":[{"id":"https://openalex.org/C44291984","wikidata":"https://www.wikidata.org/wiki/Q1074173","display_name":"Question answering","level":2,"score":0.6964258551597595},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.590765118598938},{"id":"https://openalex.org/C23123220","wikidata":"https://www.wikidata.org/wiki/Q816826","display_name":"Information retrieval","level":1,"score":0.3787664771080017},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.3754114806652069}],"mesh":[],"locations_count":2,"locations":[{"id":"doi:10.18653/v1/2025.findings-acl.1049","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-acl.1049","pdf_url":"https://aclanthology.org/2025.findings-acl.1049.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"},{"id":"pmh:oai:tubiblio.ulb.tu-darmstadt.de:155121","is_oa":false,"landing_page_url":"http://tubiblio.ulb.tu-darmstadt.de/view/person/Sachdeva=3ARachneet=3A=3A.html>","pdf_url":null,"source":{"id":"https://openalex.org/S4377196390","display_name":"TUbilio (Technical University of Darmstadt)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I31512782","host_organization_name":"Technische Universit\u00e4t Darmstadt","host_organization_lineage":["https://openalex.org/I31512782"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"acceptedVersion","is_accepted":true,"is_published":false,"raw_source_name":null,"raw_type":"Konferenzver\u00f6ffentlichung"}],"best_oa_location":{"id":"doi:10.18653/v1/2025.findings-acl.1049","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-acl.1049","pdf_url":"https://aclanthology.org/2025.findings-acl.1049.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":[{"id":"https://openalex.org/G5045554065","display_name":"Collaborative Research: RI: Medium: Multilingual Long-form QA with Retrieval-Augmented Language Models","funder_award_id":"2312949","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"}],"funders":[{"id":"https://openalex.org/F4320306076","display_name":"National Science Foundation","ror":"https://ror.org/021nxhr62"},{"id":"https://openalex.org/F4320320879","display_name":"Deutsche Forschungsgemeinschaft","ror":"https://ror.org/018mejw64"},{"id":"https://openalex.org/F4320321114","display_name":"Bundesministerium f\u00fcr Bildung und Forschung","ror":"https://ror.org/04pz7b180"}],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4412887744.pdf","grobid_xml":"https://content.openalex.org/works/W4412887744.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/W2384605597","https://openalex.org/W2387743295","https://openalex.org/W3082787378","https://openalex.org/W2136007095","https://openalex.org/W2366230879","https://openalex.org/W3208425359","https://openalex.org/W2349927912"],"abstract_inverted_index":{"Long-form":[0],"question":[1],"answering":[2],"(LFQA)":[3],"aims":[4],"to":[5,11,21,123],"provide":[6,86],"thorough":[7],"and":[8,23,43,79,85,104,132,151,163],"in-depth":[9],"answers":[10,78,138,145],"complex":[12],"questions,":[13],"enhancing":[14],"comprehension.However,":[15],"such":[16],"detailed":[17],"responses":[18],"are":[19],"prone":[20],"hallucinations":[22],"factual":[24],"inconsistencies,":[25],"challenging":[26],"their":[27],"faithful":[28],"evaluation.This":[29],"work":[30],"introduces":[31],"HaluQuestQA,":[32],"the":[33,74,119,134,137,144,157],"first":[34],"hallucination":[35],"dataset":[36,96],"with":[37,51,65,101],"localized":[38],"error":[39,54,59,99],"annotations":[40,55],"for":[41,56],"human-written":[42],"model-generated":[44],"LFQA":[45],"answers.HaluQuestQA":[46],"comprises":[47],"698":[48],"QA":[49],"pairs":[50],"1.8k":[52],"span-level":[53],"five":[57],"different":[58],"types":[60],"by":[61,147],"expert":[62],"annotators,":[63],"along":[64],"preference":[66],"judgments.Using":[67],"our":[68,148],"collected":[69],"data,":[70],"we":[71,108,128],"thoroughly":[72],"analyze":[73],"shortcomings":[75],"of":[76,136],"long-form":[77],"find":[80,143],"that":[81,97,115],"they":[82],"lack":[83],"comprehensiveness":[84],"unhelpful":[87],"references.We":[88],"train":[89],"an":[90],"automatic":[91],"feedback":[92,121],"model":[93,122],"on":[94],"this":[95],"predicts":[98],"spans":[100],"incomplete":[102],"information":[103],"provides":[105],"associated":[106],"explanations.Finally,":[107],"propose":[109],"a":[110],"promptbased":[111],"approach,":[112],"Error-Informed":[113],"Refinement,":[114],"uses":[116],"signals":[117],"from":[118],"learned":[120],"refine":[124],"generated":[125,146],"answers,":[126],"which":[127],"show":[129],"reduces":[130],"errors":[131],"improves":[133],"quality":[135],"across":[139],"multiple":[140],"models.Furthermore,":[141],"humans":[142],"approach":[149],"comprehensive":[150],"highly":[152],"prefer":[153],"them":[154],"(84%)":[155],"over":[156],"baseline":[158],"answers.":[159],"1":[160,161],"Code":[162],"data":[164],"available":[165],"at:":[166],"github.com/":[167]},"counts_by_year":[],"updated_date":"2026-07-29T14:22:42.915294","created_date":"2025-10-10T00:00:00"}