{"id":"https://openalex.org/W7225645306","doi":"https://doi.org/10.48550/arxiv.2611.08714","title":"Patching the Blind Spots: Reversible LLM Rule Overlay for Frozen Fraud Detectors","display_name":"Patching the Blind Spots: Reversible LLM Rule Overlay for Frozen Fraud Detectors","publication_year":2026,"publication_date":"2026-11-05","ids":{"openalex":"https://openalex.org/W7225645306","doi":"https://doi.org/10.48550/arxiv.2611.08714"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2611.08714","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2611.08714","pdf_url":"https://arxiv.org/pdf/2611.08714","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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"volume":null,"type":"preprint","is_traditional":true,"indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2611.08714","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5121064549","display_name":"Bo Qu","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Qu, Bo","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5161359335","display_name":"Zhurong Wang","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Zhurong","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5115002544","display_name":"Frank Zahradnik","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zahradnik, Frank","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5160210946","display_name":"Ming Ouyang","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ouyang, Ming","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":null,"last_page":null},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T12026","display_name":"Explainable Artificial Intelligence (XAI)","score":0.5005000233650208,"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":0.5005000233650208,"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/T11652","display_name":"Imbalanced Data Classification Techniques","score":0.4494999945163727,"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/T10764","display_name":"Privacy-Preserving Technologies in Data","score":0.014499999582767487,"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/gradient-boosted-decision-trees","display_name":"gradient boosting decision tree","score":0.7200000286102295},{"id":"https://openalex.org/keywords/fraud-detection","display_name":"fraud detection","score":0.7860000133514404},{"id":"https://openalex.org/keywords/shap-feature-importance","display_name":"SHAP feature importance","score":0.550000011920929},{"id":"https://openalex.org/keywords/e-commerce-fraud","display_name":"e-commerce fraud","score":0.6389999985694885},{"id":"https://openalex.org/keywords/selection-bias","display_name":"selection bias","score":0.7960000038146973},{"id":"https://openalex.org/keywords/tabular-data-classification","display_name":"tabular data classification","score":0.5360000133514404}],"concepts":[{"id":"https://openalex.org/C136085584","wikidata":"https://www.wikidata.org/wiki/Q910289","display_name":"Overlay","level":2,"score":0.7382000088691711},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6899999976158142},{"id":"https://openalex.org/C94915269","wikidata":"https://www.wikidata.org/wiki/Q1834857","display_name":"Detector","level":2,"score":0.5327000021934509},{"id":"https://openalex.org/C185798385","wikidata":"https://www.wikidata.org/wiki/Q1161707","display_name":"Benchmark (surveying)","level":2,"score":0.5282999873161316},{"id":"https://openalex.org/C100660578","wikidata":"https://www.wikidata.org/wiki/Q18733","display_name":"Recall","level":2,"score":0.5253000259399414},{"id":"https://openalex.org/C2778712577","wikidata":"https://www.wikidata.org/wiki/Q3505966","display_name":"Retraining","level":2,"score":0.5131999850273132},{"id":"https://openalex.org/C81669768","wikidata":"https://www.wikidata.org/wiki/Q2359161","display_name":"Precision and recall","level":2,"score":0.487199991941452},{"id":"https://openalex.org/C34388435","wikidata":"https://www.wikidata.org/wiki/Q2267362","display_name":"Bounded function","level":2,"score":0.4571000039577484},{"id":"https://openalex.org/C2778348673","wikidata":"https://www.wikidata.org/wiki/Q739302","display_name":"Production (economics)","level":2,"score":0.42980000376701355},{"id":"https://openalex.org/C71901391","wikidata":"https://www.wikidata.org/wiki/Q7126699","display_name":"Upload","level":2,"score":0.40619999170303345},{"id":"https://openalex.org/C106131492","wikidata":"https://www.wikidata.org/wiki/Q3072260","display_name":"Filter (signal processing)","level":2,"score":0.39820000529289246},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.36070001125335693},{"id":"https://openalex.org/C140146324","wikidata":"https://www.wikidata.org/wiki/Q1144319","display_name":"Predicate (mathematical logic)","level":2,"score":0.3571999967098236},{"id":"https://openalex.org/C2776836416","wikidata":"https://www.wikidata.org/wiki/Q1364844","display_name":"False alarm","level":2,"score":0.3513000011444092},{"id":"https://openalex.org/C2992770021","wikidata":"https://www.wikidata.org/wiki/Q7247850","display_name":"Production model","level":3,"score":0.3314000070095062},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.3278999924659729},{"id":"https://openalex.org/C79403827","wikidata":"https://www.wikidata.org/wiki/Q3988","display_name":"Real-time computing","level":1,"score":0.3221000134944916},{"id":"https://openalex.org/C99498987","wikidata":"https://www.wikidata.org/wiki/Q2210247","display_name":"Noise (video)","level":3,"score":0.3190000057220459},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.31619998812675476},{"id":"https://openalex.org/C62989814","wikidata":"https://www.wikidata.org/wiki/Q854648","display_name":"Gossip","level":2,"score":0.3003999888896942},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.27869999408721924},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.27410000562667847},{"id":"https://openalex.org/C38652104","wikidata":"https://www.wikidata.org/wiki/Q3510521","display_name":"Computer security","level":1,"score":0.26600000262260437},{"id":"https://openalex.org/C2779803651","wikidata":"https://www.wikidata.org/wiki/Q5282088","display_name":"Discriminator","level":3,"score":0.25529998540878296},{"id":"https://openalex.org/C175154964","wikidata":"https://www.wikidata.org/wiki/Q380077","display_name":"Task analysis","level":3,"score":0.2542000114917755}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2611.08714","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2611.08714","pdf_url":"https://arxiv.org/pdf/2611.08714","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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2611.08714","doi":"https://doi.org/10.48550/arxiv.2611.08714","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2611.08714","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":"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"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2611.08714","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2611.08714","pdf_url":"https://arxiv.org/pdf/2611.08714","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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7225645306.pdf","grobid_xml":"https://content.openalex.org/works/W7225645306.grobid-xml"},"referenced_works_count":2,"referenced_works":["https://openalex.org/W4414281281","https://openalex.org/W7167604932"],"related_works":["https://openalex.org/W7134971344","https://openalex.org/W7134931017","https://openalex.org/W7133832458","https://openalex.org/W7225292967","https://openalex.org/W7134951336","https://openalex.org/W7162946156","https://openalex.org/W7171295064","https://openalex.org/W4417358068","https://openalex.org/W7170090417","https://openalex.org/W7156004010"],"abstract_inverted_index":{"In":[0],"industrial":[1],"tabular":[2],"fraud":[3,103,172],"classification,":[4],"gradient-boosted":[5],"trees":[6],"remain":[7],"state-of-the-art,":[8],"but":[9],"even":[10,112],"a":[11,28,33,95,100,145,167],"well-trained":[12],"model":[13,21,34,45],"has":[14],"blind":[15],"spots,":[16],"and":[17,49,59,77,128,141,159],"retraining":[18],"the":[19,44,50,70,117,171],"deployed":[20],"is":[22,54,75,166],"not":[23],"always":[24],"safe.":[25],"We":[26],"propose":[27,39],"framework":[29],"that":[30,67],"patches":[31],"such":[32],"without":[35],"modifying":[36],"it:":[37],"LLMs":[38],"rules":[40],"anchored":[41],"on":[42,106,120,156],"features":[43],"underweights":[46],"by":[47,87],"SHAP,":[48],"resulting":[51],"noisy":[52],"pool":[53,83],"screened":[55],"for":[56,94],"empirical":[57],"support":[58],"calibrated":[60],"into":[61],"an":[62],"additive,":[63],"reversible":[64],"log-lift":[65],"overlay":[66,118],"runs":[68],"at":[69,124,131],"production":[71,136],"cut-off.":[72],"Single-LLM":[73],"benefit":[74],"high-variance":[76],"inverts":[78],"between":[79],"datasets,":[80],"so":[81],"we":[82],"three":[84],"diverse":[85],"backbones":[86],"equal-weight":[88],"logit":[89],"averaging,":[90],"trading":[91],"peak":[92],"gain":[93],"bounded":[96],"worst":[97],"case.":[98],"On":[99,144],"real-world":[101],"e-commerce":[102],"detector":[104],"trained":[105],"selection-biased":[107],"data,":[108],"continued":[109],"training":[110],"regresses":[111],"when":[113],"carefully":[114],"tuned,":[115],"whereas":[116],"gains":[119],"average":[121],"+2.26%":[122],"recall":[123,133],"equal":[125,132],"precision":[126,130,140],"($\u0394$R@P)":[127],"+1.60%":[129],"($\u0394$P@R)":[134],"over":[135],"cut-offs":[137],"spanning":[138],"30-57%":[139],"23-63%":[142],"recall.":[143],"public":[146],"benchmark":[147],"of":[148],"seven":[149],"designed":[150],"bias-shift":[151],"variants":[152],"it":[153,174],"raises":[154],"$\u0394$P@R":[155],"every":[157],"variant":[158],"keeps":[160],"$\u0394$R@P":[161],"near":[162],"neutral.":[163],"Each":[164],"amendment":[165],"readable":[168],"predicate":[169],"naming":[170],"pattern":[173],"covers.":[175]},"counts_by_year":[],"updated_date":"2026-11-12T12:28:57.059296","created_date":"2026-11-10T00:00:00"}