{"id":"https://openalex.org/W7167020386","doi":"https://doi.org/10.48550/arxiv.2607.00012","title":"PRA-RAG: Provably Robust Aggregation in Retrieval-Augmented Generation against Retrieval Corruption","display_name":"PRA-RAG: Provably Robust Aggregation in Retrieval-Augmented Generation against Retrieval Corruption","publication_year":2026,"publication_date":"2026-05-08","ids":{"openalex":"https://openalex.org/W7167020386","doi":"https://doi.org/10.48550/arxiv.2607.00012"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2607.00012","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2607.00012","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,"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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2607.00012","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5072153177","display_name":"X. Gary Tan","orcid":"https://orcid.org/0000-0003-4591-030X"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Tan, Xue","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139906231","display_name":"Yi Zheng","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zheng, Yi","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139868276","display_name":"Chang Huo","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Huo, Chang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5059740305","display_name":"Yunruo Zhang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Yunruo","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139903529","display_name":"Yu Liu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Liu, Yu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5045726084","display_name":"Hao Luan","orcid":"https://orcid.org/0000-0001-6885-1504"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Luan, Hao","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5132702301","display_name":"Zhuyang Yu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yu, Zhuyang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139936367","display_name":"Xiaoyan Sun","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Sun, Xiaoyan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139892412","display_name":"Ping Chen","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chen, Ping","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5139885104","display_name":"Jun Dai","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Dai, Jun","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":false,"cited_by_count":0,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":null,"last_page":null},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.715499997138977,"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":0.715499997138977,"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/T11273","display_name":"Advanced Graph Neural Networks","score":0.06509999930858612,"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/T11714","display_name":"Multimodal Machine Learning Applications","score":0.04540000110864639,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"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/robustness","display_name":"Robustness (evolution)","score":0.6948999762535095},{"id":"https://openalex.org/keywords/embedding","display_name":"Embedding","score":0.5741999745368958},{"id":"https://openalex.org/keywords/representation","display_name":"Representation (politics)","score":0.48249998688697815},{"id":"https://openalex.org/keywords/measure","display_name":"Measure (data warehouse)","score":0.430400013923645},{"id":"https://openalex.org/keywords/key","display_name":"Key (lock)","score":0.296999990940094}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7590000033378601},{"id":"https://openalex.org/C63479239","wikidata":"https://www.wikidata.org/wiki/Q7353546","display_name":"Robustness (evolution)","level":3,"score":0.6948999762535095},{"id":"https://openalex.org/C41608201","wikidata":"https://www.wikidata.org/wiki/Q980509","display_name":"Embedding","level":2,"score":0.5741999745368958},{"id":"https://openalex.org/C2776359362","wikidata":"https://www.wikidata.org/wiki/Q2145286","display_name":"Representation (politics)","level":3,"score":0.48249998688697815},{"id":"https://openalex.org/C2780009758","wikidata":"https://www.wikidata.org/wiki/Q6804172","display_name":"Measure (data warehouse)","level":2,"score":0.430400013923645},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3962000012397766},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.3416999876499176},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.3034000098705292},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.2976999878883362},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.296999990940094},{"id":"https://openalex.org/C2778572836","wikidata":"https://www.wikidata.org/wiki/Q380933","display_name":"Space (punctuation)","level":2,"score":0.25870001316070557},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.25450000166893005},{"id":"https://openalex.org/C122770356","wikidata":"https://www.wikidata.org/wiki/Q1656753","display_name":"Identifiability","level":2,"score":0.2535000145435333},{"id":"https://openalex.org/C67226441","wikidata":"https://www.wikidata.org/wiki/Q1665389","display_name":"Robust statistics","level":3,"score":0.2513999938964844}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2607.00012","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2607.00012","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,"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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2607.00012","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2607.00012","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,"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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"},"sustainable_development_goals":[{"display_name":"Peace, Justice and strong institutions","id":"https://metadata.un.org/sdg/16","score":0.8042623996734619}],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Retrieval-Augmented":[0],"Generation":[1],"(RAG)":[2],"enhances":[3],"Large":[4],"Language":[5],"Models":[6],"(LLMs)":[7],"by":[8],"incorporating":[9],"external":[10],"knowledge,":[11],"effectively":[12],"mitigating":[13],"their":[14],"inherent":[15],"knowledge":[16,49],"limitations.":[17],"However,":[18],"RAG":[19,128],"remains":[20],"vulnerable":[21],"to":[22,29,67,90,138],"poisoning":[23,70],"attacks":[24,71],"that":[25,131],"manipulate":[26],"retrieved":[27,52,73,80,113],"texts":[28,81],"mislead":[30],"model":[31],"outputs.":[32],"Existing":[33],"defense":[34],"mechanisms":[35],"often":[36],"lack":[37],"theoretical":[38,105],"robustness":[39],"guarantees":[40],"and":[41,82,115,127],"perform":[42],"unreliably":[43],"when":[44],"the":[45,51,87,108,134],"LLM":[46],"has":[47],"limited":[48],"of":[50,79,111,120,147],"content.":[53],"In":[54],"this":[55],"work,":[56],"we":[57],"propose":[58],"PRA-RAG,":[59],"a":[60,92,97,117],"provably":[61],"robust":[62,93],"retrieval":[63],"aggregation":[64],"algorithm":[65],"designed":[66],"defend":[68],"against":[69],"on":[72,107],"texts.":[74],"PRA-RAG":[75,132],"samples":[76],"multiple":[77,125],"combinations":[78],"utilizes":[83],"geometric":[84],"structures":[85],"in":[86],"embedding":[88],"space":[89],"identify":[91],"subset,":[94],"from":[95],"which":[96],"stable":[98],"aggregated":[99],"representation":[100],"is":[101],"derived.":[102],"We":[103],"provide":[104],"bounds":[106],"maximum":[109],"impact":[110],"poisoned":[112],"content":[114],"establish":[116],"quantitative":[118],"measure":[119],"RAG's":[121],"robustness.":[122],"Experiments":[123],"across":[124],"benchmarks":[126],"architectures":[129],"demonstrate":[130],"reduces":[133],"attack":[135],"success":[136],"rate":[137],"as":[139,141],"low":[140],"1%":[142],"while":[143],"maintaining":[144],"an":[145],"accuracy":[146],"71%,":[148],"significantly":[149],"outperforming":[150],"representative":[151],"state-of-the-art":[152],"methods.":[153]},"counts_by_year":[],"updated_date":"2026-07-28T07:46:37.118299","created_date":"2026-07-03T00:00:00"}