{"id":"https://openalex.org/W4412945043","doi":"https://doi.org/10.18653/v1/2025.acl-long.869","title":"Masking in Multi-hop QA: An Analysis of How Language Models Perform with Context Permutation","display_name":"Masking in Multi-hop QA: An Analysis of How Language Models Perform with Context Permutation","publication_year":2025,"publication_date":"2025-01-01","ids":{"openalex":"https://openalex.org/W4412945043","doi":"https://doi.org/10.18653/v1/2025.acl-long.869"},"language":"en","primary_location":{"id":"doi:10.18653/v1/2025.acl-long.869","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.acl-long.869","pdf_url":"https://aclanthology.org/2025.acl-long.869.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":"Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2025.acl-long.869.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5032466487","display_name":"Wenyu Huang","orcid":"https://orcid.org/0000-0001-9722-5502"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wenyu Huang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5061611097","display_name":"Pavlos Vougiouklis","orcid":"https://orcid.org/0000-0002-1781-6319"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Pavlos Vougiouklis","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5041024491","display_name":"Mirella Lapata","orcid":"https://orcid.org/0000-0002-2107-1516"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Mirella Lapata","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5023993463","display_name":"Jeff Z. Pan","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jeff Z. Pan","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":null,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"17781","last_page":"17795"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10201","display_name":"Speech Recognition and Synthesis","score":0.989300012588501,"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/T10201","display_name":"Speech Recognition and Synthesis","score":0.989300012588501,"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/T11309","display_name":"Music and Audio Processing","score":0.9596999883651733,"subfield":{"id":"https://openalex.org/subfields/1711","display_name":"Signal Processing"},"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/T10181","display_name":"Natural Language Processing Techniques","score":0.9355999827384949,"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":0.7909201383590698},{"id":"https://openalex.org/keywords/permutation","display_name":"Permutation (music)","score":0.516913890838623},{"id":"https://openalex.org/keywords/context","display_name":"Context (archaeology)","score":0.49270299077033997},{"id":"https://openalex.org/keywords/masking","display_name":"Masking (illustration)","score":0.46152180433273315},{"id":"https://openalex.org/keywords/natural-language-processing","display_name":"Natural language processing","score":0.43683648109436035},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.3787202537059784}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7909201383590698},{"id":"https://openalex.org/C21308566","wikidata":"https://www.wikidata.org/wiki/Q7169365","display_name":"Permutation (music)","level":2,"score":0.516913890838623},{"id":"https://openalex.org/C2779343474","wikidata":"https://www.wikidata.org/wiki/Q3109175","display_name":"Context (archaeology)","level":2,"score":0.49270299077033997},{"id":"https://openalex.org/C2777402240","wikidata":"https://www.wikidata.org/wiki/Q6783436","display_name":"Masking (illustration)","level":2,"score":0.46152180433273315},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.43683648109436035},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3787202537059784},{"id":"https://openalex.org/C121332964","wikidata":"https://www.wikidata.org/wiki/Q413","display_name":"Physics","level":0,"score":0.0},{"id":"https://openalex.org/C151730666","wikidata":"https://www.wikidata.org/wiki/Q7205","display_name":"Paleontology","level":1,"score":0.0},{"id":"https://openalex.org/C153349607","wikidata":"https://www.wikidata.org/wiki/Q36649","display_name":"Visual arts","level":1,"score":0.0},{"id":"https://openalex.org/C142362112","wikidata":"https://www.wikidata.org/wiki/Q735","display_name":"Art","level":0,"score":0.0},{"id":"https://openalex.org/C24890656","wikidata":"https://www.wikidata.org/wiki/Q82811","display_name":"Acoustics","level":1,"score":0.0},{"id":"https://openalex.org/C86803240","wikidata":"https://www.wikidata.org/wiki/Q420","display_name":"Biology","level":0,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2025.acl-long.869","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.acl-long.869","pdf_url":"https://aclanthology.org/2025.acl-long.869.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":"Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2025.acl-long.869","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.acl-long.869","pdf_url":"https://aclanthology.org/2025.acl-long.869.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":"Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},"sustainable_development_goals":[{"id":"https://metadata.un.org/sdg/10","score":0.41999998688697815,"display_name":"Reduced inequalities"}],"awards":[{"id":"https://openalex.org/G373773752","display_name":null,"funder_award_id":"EP/S022481/1","funder_id":"https://openalex.org/F4320314731","funder_display_name":"UK Research and Innovation"}],"funders":[{"id":"https://openalex.org/F4320314731","display_name":"UK Research and Innovation","ror":"https://ror.org/001aqnf71"},{"id":"https://openalex.org/F4320320332","display_name":"University of Edinburgh","ror":"https://ror.org/01nrxwf90"}],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4412945043.pdf","grobid_xml":"https://content.openalex.org/works/W4412945043.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W3081694532","https://openalex.org/W1969211203","https://openalex.org/W4411737141","https://openalex.org/W1517958729","https://openalex.org/W2092272653","https://openalex.org/W1992704972","https://openalex.org/W271627879","https://openalex.org/W2594778474","https://openalex.org/W2349444258","https://openalex.org/W3204019825"],"abstract_inverted_index":{"Multi-hop":[0],"Question":[1],"Answering":[2],"(MHQA)":[3],"adds":[4],"layers":[5],"of":[6,115,172,175,182],"complexity":[7],"to":[8,56,68,164,190,204],"question":[9],"answering,":[10],"making":[11],"it":[12],"more":[13],"challenging.When":[14],"Language":[15],"Models":[16],"(LMs)":[17],"are":[18,25],"prompted":[19],"with":[20,29,130,139,149],"multiple":[21],"search":[22,73],"results,":[23],"they":[24],"tasked":[26],"not":[27],"only":[28],"retrieving":[30],"relevant":[31],"information":[32,40],"but":[33],"also":[34],"employing":[35],"multi-hop":[36,69],"reasoning":[37,141],"across":[38,58],"the":[39,49,91,94,113,135,140,154,165,173,180,196],"sources.Although":[41],"LMs":[42,66,101],"perform":[43],"well":[44],"on":[45,209],"traditional":[46],"question-answering":[47],"tasks,":[48,104],"causal":[50,99,146,155],"mask":[51,156],"can":[52,157],"hinder":[53],"their":[54,160],"capacity":[55],"reason":[57],"complex":[59],"contexts.In":[60],"this":[61,202,210],"paper,":[62],"we":[63,167],"explore":[64],"how":[65],"respond":[67],"questions":[70],"by":[71,152],"permuting":[72],"results":[74],"(retrieved":[75],"documents)":[76],"under":[77],"various":[78],"configurations.Our":[79],"study":[80],"reveals":[81,118],"interesting":[82],"findings":[83],"as":[84,90],"follows:":[85],"1)":[86],"Encoderdecoder":[87],"models,":[88,129],"such":[89],"ones":[92],"in":[93,102,109,121,179],"Flan-T5":[95],"family,":[96],"generally":[97],"outperform":[98],"decoderonly":[100,128],"MHQA":[103],"despite":[105],"being":[106],"significantly":[107],"smaller":[108],"size;":[110],"2)":[111],"altering":[112],"order":[114,137],"gold":[116],"documents":[117],"distinct":[119],"trends":[120],"both":[122],"Flan":[123],"T5":[124],"models":[125,148],"and":[126],"fine-tuned":[127],"optimal":[131],"performance":[132,208],"observed":[133],"when":[134,195],"document":[136],"aligns":[138],"chain":[142],"order;":[143],"3)":[144],"enhancing":[145],"decoder-only":[147],"bi-directional":[150],"attention":[151,177,187],"modifying":[153],"effectively":[158],"boost":[159],"end":[161],"performance.In":[162],"addition":[163],"above,":[166],"conduct":[168],"a":[169],"thorough":[170],"investigation":[171],"distribution":[174],"LM":[176],"weights":[178,188],"context":[181],"MHQA.Our":[183],"experiments":[184],"reveal":[185],"that":[186],"tend":[189],"peak":[191],"at":[192,216],"higher":[193],"values":[194],"resulting":[197],"answer":[198],"is":[199,213],"correct.We":[200],"leverage":[201],"finding":[203],"heuristically":[205],"improve":[206],"LMs'":[207],"task.Our":[211],"code":[212],"publicly":[214],"available":[215],"https://github.":[217],"com/hwy9855/MultiHopQA-Reasoning.":[218]},"counts_by_year":[],"updated_date":"2026-07-14T23:27:15.235271","created_date":"2025-10-10T00:00:00"}