{"id":"https://openalex.org/W7160870304","doi":"https://doi.org/10.48550/arxiv.2605.07527","title":"Why Self-Inconsistency Arises in GNN Explanations and How to Exploit It","display_name":"Why Self-Inconsistency Arises in GNN Explanations and How to Exploit It","publication_year":2026,"publication_date":"2026-05-08","ids":{"openalex":"https://openalex.org/W7160870304","doi":"https://doi.org/10.48550/arxiv.2605.07527"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2605.07527","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.07527","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"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":"article"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2605.07527","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5090741164","display_name":"Wenxin Tai","orcid":"https://orcid.org/0000-0001-7364-8324"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Tai, Wenxin","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5135861508","display_name":"Yaqian Liu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Liu, Yaqian","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5034789908","display_name":"Ting Zhong","orcid":"https://orcid.org/0000-0002-8163-3146"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhong, Ting","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5135873140","display_name":"Fan Zhou","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhou, Fan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":4,"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/T12026","display_name":"Explainable Artificial Intelligence (XAI)","score":0.8669999837875366,"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":0.8669999837875366,"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":0.0934000015258789,"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/T13702","display_name":"Machine Learning in Healthcare","score":0.007499999832361937,"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/exploit","display_name":"Exploit","score":0.7455999851226807},{"id":"https://openalex.org/keywords/graph","display_name":"Graph","score":0.4713999927043915},{"id":"https://openalex.org/keywords/regularization","display_name":"Regularization (linguistics)","score":0.4424000084400177},{"id":"https://openalex.org/keywords/context","display_name":"Context (archaeology)","score":0.4124999940395355},{"id":"https://openalex.org/keywords/deep-neural-networks","display_name":"Deep neural networks","score":0.33309999108314514},{"id":"https://openalex.org/keywords/perturbation","display_name":"Perturbation (astronomy)","score":0.3037000000476837}],"concepts":[{"id":"https://openalex.org/C165696696","wikidata":"https://www.wikidata.org/wiki/Q11287","display_name":"Exploit","level":2,"score":0.7455999851226807},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.620199978351593},{"id":"https://openalex.org/C132525143","wikidata":"https://www.wikidata.org/wiki/Q141488","display_name":"Graph","level":2,"score":0.4713999927043915},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.4668999910354614},{"id":"https://openalex.org/C2776135515","wikidata":"https://www.wikidata.org/wiki/Q17143721","display_name":"Regularization (linguistics)","level":2,"score":0.4424000084400177},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4401000142097473},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.4138000011444092},{"id":"https://openalex.org/C2779343474","wikidata":"https://www.wikidata.org/wiki/Q3109175","display_name":"Context (archaeology)","level":2,"score":0.4124999940395355},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.4020000100135803},{"id":"https://openalex.org/C2984842247","wikidata":"https://www.wikidata.org/wiki/Q197536","display_name":"Deep neural networks","level":3,"score":0.33309999108314514},{"id":"https://openalex.org/C177918212","wikidata":"https://www.wikidata.org/wiki/Q803623","display_name":"Perturbation (astronomy)","level":2,"score":0.3037000000476837},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.29190000891685486},{"id":"https://openalex.org/C50644808","wikidata":"https://www.wikidata.org/wiki/Q192776","display_name":"Artificial neural network","level":2,"score":0.2806999981403351},{"id":"https://openalex.org/C2779960059","wikidata":"https://www.wikidata.org/wiki/Q7113681","display_name":"Overhead (engineering)","level":2,"score":0.2754000127315521},{"id":"https://openalex.org/C2779530757","wikidata":"https://www.wikidata.org/wiki/Q1207505","display_name":"Quality (philosophy)","level":2,"score":0.27000001072883606},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.2619999945163727},{"id":"https://openalex.org/C88230418","wikidata":"https://www.wikidata.org/wiki/Q131476","display_name":"Graph theory","level":2,"score":0.2572999894618988},{"id":"https://openalex.org/C185798385","wikidata":"https://www.wikidata.org/wiki/Q1161707","display_name":"Benchmark (surveying)","level":2,"score":0.2533000111579895}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2605.07527","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.07527","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"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":"article"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2605.07527","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.07527","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"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":"article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Recent":[0],"work":[1],"has":[2],"observed":[3],"that":[4,85,107,131],"explanations":[5,109],"produced":[6],"by":[7],"Self-Interpretable":[8],"Graph":[9],"Neural":[10],"Networks":[11],"(SI-GNNs)":[12],"can":[13],"be":[14],"self-inconsistent:":[15],"when":[16],"the":[17,50,94],"model":[18],"is":[19],"reapplied":[20],"to":[21,64,72],"its":[22],"own":[23],"explanatory":[24],"graph":[25],"subset,":[26],"it":[27],"may":[28],"produce":[29],"a":[30,59,101],"different":[31],"explanation.":[32],"However,":[33],"why":[34,66],"self-inconsistency":[35],"arises":[36],"remains":[37],"poorly":[38],"understood.":[39],"In":[40],"this":[41,73],"work,":[42],"we":[43,97],"first":[44],"identify":[45],"re-explanation-induced":[46],"context":[47],"perturbation":[48],"as":[49],"direct":[51],"cause":[52],"of":[53],"score":[54],"variation.":[55],"We":[56],"then":[57],"introduce":[58],"latent":[60,81],"signal":[61,82],"assignment":[62],"hypothesis":[63,128],"explain":[65],"only":[67,111,139],"some":[68],"edges":[69,87],"are":[70],"sensitive":[71],"perturbation,":[74],"and":[75,103,123,129],"analyze":[76],"how":[77],"conciseness":[78],"regularization":[79],"affects":[80],"assignment.":[83],"Given":[84],"self-inconsistent":[86],"do":[88],"not":[89],"provide":[90],"stable":[91],"evidence":[92],"for":[93],"model's":[95],"prediction,":[96],"propose":[98],"Self-Denoising":[99],"(SD),":[100],"model-agnostic":[102],"training-free":[104],"post-processing":[105],"strategy":[106],"calibrates":[108],"with":[110],"one":[112],"additional":[113],"forward":[114],"pass.":[115],"Experiments":[116],"across":[117],"representative":[118],"SI-GNN":[119],"frameworks,":[120],"backbone":[121],"architectures,":[122],"benchmark":[124],"datasets":[125],"support":[126],"our":[127],"show":[130],"SD":[132],"consistently":[133],"improves":[134],"explanation":[135],"quality":[136],"while":[137],"adding":[138],"about":[140],"4--6\\%":[141],"computational":[142],"overhead":[143],"in":[144],"practice.":[145]},"counts_by_year":[],"updated_date":"2026-06-11T09:08:48.828518","created_date":"2026-05-12T00:00:00"}