{"id":"https://openalex.org/W4412888287","doi":"https://doi.org/10.18653/v1/2025.findings-acl.530","title":"MultiHoax: A Dataset of Multi-hop False-premise questions","display_name":"MultiHoax: A Dataset of Multi-hop False-premise questions","publication_year":2025,"publication_date":"2025-01-01","ids":{"openalex":"https://openalex.org/W4412888287","doi":"https://doi.org/10.18653/v1/2025.findings-acl.530"},"language":"en","primary_location":{"id":"doi:10.18653/v1/2025.findings-acl.530","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-acl.530","pdf_url":"https://aclanthology.org/2025.findings-acl.530.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.530.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5113253015","display_name":"Mohammadamin Shafiei","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Mohammadamin Shafiei","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5099137350","display_name":"Hamidreza Saffari","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Hamidreza Saffari","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5054918343","display_name":"Nafise Sadat Moosavi","orcid":"https://orcid.org/0000-0002-8332-307X"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Nafise Sadat Moosavi","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":0.14335241,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"10169","last_page":"10187"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10181","display_name":"Natural Language Processing Techniques","score":0.9751999974250793,"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/T10181","display_name":"Natural Language Processing Techniques","score":0.9751999974250793,"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/premise","display_name":"Premise","score":0.7864181995391846},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.6974111199378967},{"id":"https://openalex.org/keywords/hop","display_name":"Hop (telecommunications)","score":0.5774688124656677},{"id":"https://openalex.org/keywords/data-science","display_name":"Data science","score":0.3464117646217346},{"id":"https://openalex.org/keywords/computer-network","display_name":"Computer network","score":0.1838148534297943},{"id":"https://openalex.org/keywords/epistemology","display_name":"Epistemology","score":0.06454017758369446}],"concepts":[{"id":"https://openalex.org/C2778023277","wikidata":"https://www.wikidata.org/wiki/Q321703","display_name":"Premise","level":2,"score":0.7864181995391846},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6974111199378967},{"id":"https://openalex.org/C25906391","wikidata":"https://www.wikidata.org/wiki/Q1432381","display_name":"Hop (telecommunications)","level":2,"score":0.5774688124656677},{"id":"https://openalex.org/C2522767166","wikidata":"https://www.wikidata.org/wiki/Q2374463","display_name":"Data science","level":1,"score":0.3464117646217346},{"id":"https://openalex.org/C31258907","wikidata":"https://www.wikidata.org/wiki/Q1301371","display_name":"Computer network","level":1,"score":0.1838148534297943},{"id":"https://openalex.org/C111472728","wikidata":"https://www.wikidata.org/wiki/Q9471","display_name":"Epistemology","level":1,"score":0.06454017758369446},{"id":"https://openalex.org/C138885662","wikidata":"https://www.wikidata.org/wiki/Q5891","display_name":"Philosophy","level":0,"score":0.0}],"mesh":[],"locations_count":2,"locations":[{"id":"doi:10.18653/v1/2025.findings-acl.530","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-acl.530","pdf_url":"https://aclanthology.org/2025.findings-acl.530.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:eprints.whiterose.ac.uk:241587","is_oa":false,"landing_page_url":"https://orcid.org/0000-0002-8332-307X>","pdf_url":null,"source":{"id":"https://openalex.org/S4306400854","display_name":"White Rose Research Online (University of Leeds, The University of Sheffield, University of York)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I2800616092","host_organization_name":"White Rose University Consortium","host_organization_lineage":["https://openalex.org/I2800616092"],"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":"Proceedings Paper"}],"best_oa_location":{"id":"doi:10.18653/v1/2025.findings-acl.530","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-acl.530","pdf_url":"https://aclanthology.org/2025.findings-acl.530.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":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4412888287.pdf","grobid_xml":"https://content.openalex.org/works/W4412888287.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/W590788508","https://openalex.org/W4235873430","https://openalex.org/W2611974471","https://openalex.org/W4313233093","https://openalex.org/W2358082531","https://openalex.org/W2589976903","https://openalex.org/W2335596023"],"abstract_inverted_index":{"As":[0],"Large":[1],"Language":[2],"Models":[3],"are":[4],"increasingly":[5],"deployed":[6],"in":[7,86,145],"high-stakes":[8],"domains,":[9],"their":[10],"ability":[11,81],"to":[12,40,82,107,118],"detect":[13,119],"false":[14,84,120,136],"assumptions":[15,38],"and":[16,95,127,139],"reason":[17],"critically":[18],"is":[19],"crucial":[20],"for":[21,78,134],"ensuring":[22],"reliable":[23],"outputs.False-premise":[24],"questions":[25],"(FPQs)":[26],"serve":[27],"as":[28,102],"an":[29],"important":[30],"evaluation":[31],"method":[32],"by":[33],"exposing":[34],"cases":[35],"where":[36,55],"flawed":[37],"lead":[39],"incorrect":[41],"responses.While":[42],"existing":[43],"benchmarks":[44],"focus":[45],"on":[46,67],"single-hop":[47],"FPQs,":[48],"real-world":[49],"reasoning":[50,62,89,110,129,143],"often":[51],"requires":[52],"multi-hop":[53,128,142],"inference,":[54],"models":[56],"must":[57],"verify":[58],"consistency":[59],"across":[60,111,122],"multiple":[61],"steps":[63],"rather":[64],"than":[65],"relying":[66],"surface-level":[68],"cues.To":[69],"address":[70],"this":[71],"gap,":[72],"we":[73],"introduce":[74],"MultiHoax,":[75],"a":[76],"benchmark":[77],"evaluating":[79],"LLMs'":[80],"handle":[83],"premises":[85,121],"complex,":[87],"multi-step":[88],"tasks.Our":[90],"dataset":[91],"spans":[92],"seven":[93],"countries":[94],"ten":[96],"diverse":[97],"knowledge":[98,105,125],"categories,":[99,126],"using":[100],"Wikipedia":[101],"the":[103,132],"primary":[104],"source":[106],"enable":[108],"factual":[109],"regions.Experiments":[112],"reveal":[113],"that":[114],"state-of-the-art":[115],"LLMs":[116],"struggle":[117],"different":[123],"countries,":[124],"types,":[130],"highlighting":[131],"need":[133],"improved":[135],"premise":[137],"detection":[138],"more":[140],"robust":[141],"capabilities":[144],"LLMs.":[146]},"counts_by_year":[],"updated_date":"2026-07-29T14:22:42.915294","created_date":"2025-10-10T00:00:00"}