{"id":"https://openalex.org/W7140092366","doi":"https://doi.org/10.18653/v1/2026.findings-eacl.310","title":"Argument-Based Consistency in Toxicity Explanations of LLMs","display_name":"Argument-Based Consistency in Toxicity Explanations of LLMs","publication_year":2026,"publication_date":"2026-01-01","ids":{"openalex":"https://openalex.org/W7140092366","doi":"https://doi.org/10.18653/v1/2026.findings-eacl.310"},"language":"en","primary_location":{"id":"doi:10.18653/v1/2026.findings-eacl.310","doi":"https://doi.org/10.18653/v1/2026.findings-eacl.310","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-eacl.310","pdf_url":null,"source":{"id":"https://openalex.org/S7407087051","display_name":"Findings of the Association for Computational Linguistics: EACL","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"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: EACL 2026","raw_type":"proceedings-article"},"volume":null,"type":"conference-paper","indexed_in":["arxiv","crossref","datacite"],"open_access":{"is_oa":true,"oa_status":"hybrid","oa_url":"https://doi.org/10.18653/v1/2026.findings-eacl.310","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5029382491","display_name":"Ramaravind Kommiya Mothilal","orcid":"https://orcid.org/0000-0002-6469-3091","observed_orcids":["https://orcid.org/0000-0002-6469-3091"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ramaravind Kommiya Mothilal","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5126067198","display_name":"Joanna Roy","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Joanna Roy","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5130344979","display_name":"Syed Ishtiaque Ahmed","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Syed Ishtiaque Ahmed","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5100659941","display_name":"Shion Guha","orcid":"https://orcid.org/0000-0003-0073-2378","observed_orcids":["https://orcid.org/0000-0003-0073-2378"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Shion Guha","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":17.0938,"has_fulltext":false,"cited_by_count":2,"citation_normalized_percentile":{"value":0.98777697,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"cited_by_percentile_year":{"min":96,"max":97},"biblio":{"volume":null,"issue":null,"first_page":"5913","last_page":"5941"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T12026","display_name":"Explainable Artificial Intelligence (XAI)","score":0.9384999871253967,"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/T12026","display_name":"Explainable Artificial Intelligence (XAI)","score":0.9384999871253967,"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/T10883","display_name":"Ethics and Social Impacts of AI","score":0.01510000042617321,"subfield":{"id":"https://openalex.org/subfields/3311","display_name":"Safety Research"},"field":{"id":"https://openalex.org/fields/33","display_name":"Social Sciences"},"domain":{"id":"https://openalex.org/domains/2","display_name":"Social Sciences"}},{"id":"https://openalex.org/T13295","display_name":"Safety Systems Engineering in Autonomy","score":0.007600000128149986,"subfield":{"id":"https://openalex.org/subfields/2213","display_name":"Safety, Risk, Reliability and Quality"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/large-language-models","display_name":"large language models","score":0.871999979019165},{"id":"https://openalex.org/keywords/toxicity","display_name":"toxicity","score":0.6069999933242798},{"id":"https://openalex.org/keywords/explanation-consistency","display_name":"explanation consistency","score":0.6190000176429749},{"id":"https://openalex.org/keywords/argument-based-reasoning","display_name":"argument-based reasoning","score":0.75}],"concepts":[{"id":"https://openalex.org/C2776436953","wikidata":"https://www.wikidata.org/wiki/Q5163215","display_name":"Consistency (knowledge bases)","level":2,"score":0.6601999998092651},{"id":"https://openalex.org/C71924100","wikidata":"https://www.wikidata.org/wiki/Q11190","display_name":"Medicine","level":0,"score":0.4108000099658966},{"id":"https://openalex.org/C29730261","wikidata":"https://www.wikidata.org/wiki/Q274160","display_name":"Toxicity","level":2,"score":0.40059998631477356},{"id":"https://openalex.org/C99454951","wikidata":"https://www.wikidata.org/wiki/Q932068","display_name":"Environmental health","level":1,"score":0.37709999084472656},{"id":"https://openalex.org/C12174686","wikidata":"https://www.wikidata.org/wiki/Q1058438","display_name":"Risk assessment","level":2,"score":0.34689998626708984},{"id":"https://openalex.org/C15744967","wikidata":"https://www.wikidata.org/wiki/Q9418","display_name":"Psychology","level":0,"score":0.30799999833106995},{"id":"https://openalex.org/C162118730","wikidata":"https://www.wikidata.org/wiki/Q1128453","display_name":"Actuarial science","level":1,"score":0.30559998750686646},{"id":"https://openalex.org/C77805123","wikidata":"https://www.wikidata.org/wiki/Q161272","display_name":"Social psychology","level":1,"score":0.30000001192092896},{"id":"https://openalex.org/C64357122","wikidata":"https://www.wikidata.org/wiki/Q1149766","display_name":"Causality (physics)","level":2,"score":0.29789999127388},{"id":"https://openalex.org/C2776035688","wikidata":"https://www.wikidata.org/wiki/Q1606558","display_name":"Affect (linguistics)","level":2,"score":0.2849000096321106},{"id":"https://openalex.org/C112930515","wikidata":"https://www.wikidata.org/wiki/Q4389547","display_name":"Risk analysis (engineering)","level":1,"score":0.2831000089645386},{"id":"https://openalex.org/C2780791683","wikidata":"https://www.wikidata.org/wiki/Q846785","display_name":"Action (physics)","level":2,"score":0.27570000290870667}],"mesh":[],"locations_count":3,"locations":[{"id":"doi:10.18653/v1/2026.findings-eacl.310","doi":"https://doi.org/10.18653/v1/2026.findings-eacl.310","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-eacl.310","pdf_url":null,"source":{"id":"https://openalex.org/S7407087051","display_name":"Findings of the Association for Computational Linguistics: EACL","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"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: EACL 2026","raw_type":"proceedings-article"},{"id":"pmh:oai:arXiv.org:2506.19113","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2506.19113","pdf_url":"https://arxiv.org/pdf/2506.19113","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2506.19113","doi":"https://doi.org/10.48550/arxiv.2506.19113","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2506.19113","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.18653/v1/2026.findings-eacl.310","doi":"https://doi.org/10.18653/v1/2026.findings-eacl.310","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-eacl.310","pdf_url":null,"source":{"id":"https://openalex.org/S7407087051","display_name":"Findings of the Association for Computational Linguistics: EACL","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"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: EACL 2026","raw_type":"proceedings-article"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W7153157207","https://openalex.org/W4404781097","https://openalex.org/W7219431830","https://openalex.org/W7092708227","https://openalex.org/W7125580424","https://openalex.org/W7172404594","https://openalex.org/W7163007493","https://openalex.org/W4402670855","https://openalex.org/W4415798831","https://openalex.org/W7133228402"],"abstract_inverted_index":{"The":[0],"discourse":[1],"around":[2,10],"toxicity":[3,23,56,91,117,140,157,177],"and":[4,96,132,175,182,190],"LLMs":[5,147],"in":[6,37,115,180],"NLP":[7],"largely":[8],"revolves":[9],"detection":[11],"tasks.":[12,39],"This":[13],"work":[14],"shifts":[15],"the":[16,85,113,163,167,172],"focus":[17],"to":[18,33,49,53,59,87,110,130,151],"evaluating":[19],"LLMs'":[20,89,116],"reasoning":[21,155],"about":[22,156,162],"-":[24,32],"from":[25],"their":[26,35,60,154,176],"explanations":[27,92,150,192],"that":[28,83,145],"justify":[29],"a":[30,75],"stance":[31],"enhance":[34],"trustworthiness":[36],"downstream":[38],"Despite":[40],"extensive":[41],"research":[42],"on":[43,62,101,123,137],"explainability,":[44],"it":[45],"is":[46],"not":[47],"straightforward":[48],"adopt":[50],"existing":[51],"methods":[52],"evaluate":[54,112],"free-form":[55,90],"explanation":[57],"due":[58],"over-reliance":[61],"input":[63],"text":[64],"perturbations,":[65],"among":[66],"other":[67],"challenges.":[68],"To":[69],"account":[70],"for":[71,108,194],"these,":[72],"we":[73,104],"propose":[74],"novel,":[76],"theoretically-grounded":[77],"multi-dimensional":[78],"criterion,":[79],"Argument-based":[80],"Consistency":[81],"(ArC),":[82],"measures":[84],"extent":[86],"which":[88],"reflect":[93],"an":[94,133],"ideal":[95],"logical":[97],"argumentation":[98],"process.":[99],"Based":[100],"uncertainty":[102],"quantification,":[103],"develop":[105],"six":[106],"metrics":[107],"ArC":[109],"comprehensively":[111],"(in)consistencies":[114],"explanations.":[118],"We":[119,185],"conduct":[120],"several":[121],"experiments":[122],"three":[124],"Llama":[125],"models":[126],"(of":[127],"size":[128],"up":[129],"70B)":[131],"8B":[134],"Ministral":[135],"model":[136],"five":[138],"diverse":[139],"datasets.":[141],"Our":[142],"results":[143],"show":[144],"while":[146],"generate":[148],"plausible":[149],"simple":[152],"prompts,":[153],"breaks":[158],"down":[159],"when":[160],"prompted":[161],"nuanced":[164],"relations":[165],"between":[166],"complete":[168],"set":[169],"of":[170],"reasons,":[171,174],"individual":[173],"stances,":[178],"resulting":[179],"inconsistent":[181],"irrelevant":[183],"responses.":[184],"open-source":[186],"our":[187],"code":[188],"(https://github.com/uofthcdslab/ArC)":[189],"LLM-generated":[191],"(https://huggingface.co/collections/uofthcdslab/arc)":[193],"future":[195],"works.":[196]},"counts_by_year":[{"year":2026,"cited_by_count":2}],"updated_date":"2026-10-12T09:40:04.860316","created_date":"2026-03-24T00:00:00"}