{"id":"https://openalex.org/W7128508703","doi":"https://doi.org/10.48550/arxiv.2602.08500","title":"Is Meta-Path Attention an Explanation? Evidence of Alignment and Decoupling in Heterogeneous GNNs","display_name":"Is Meta-Path Attention an Explanation? Evidence of Alignment and Decoupling in Heterogeneous GNNs","publication_year":2026,"publication_date":"2026-02-09","ids":{"openalex":"https://openalex.org/W7128508703","doi":"https://doi.org/10.48550/arxiv.2602.08500"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2602.08500","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2602.08500","pdf_url":"https://arxiv.org/pdf/2602.08500","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"},"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/2602.08500","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5070537286","display_name":"Maiqi Jiang","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jiang, Maiqi","raw_affiliation_strings":["Williamsburg , Virginia , USA"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Williamsburg , Virginia , USA","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5125561846","display_name":"Noman Ali","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I154549908","display_name":"Indian Institute of Technology Jodhpur","ror":"https://ror.org/03yacj906","country_code":"IN","type":"education","lineage":["https://openalex.org/I154549908"]}],"countries":["IN"],"is_corresponding":false,"raw_author_name":"Ali, Noman","raw_affiliation_strings":["Indian Institute of Technology Jodhpur , India"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Indian Institute of Technology Jodhpur , India","institution_ids":["https://openalex.org/I154549908"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5125557486","display_name":"Yiran Ding","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I27804330","display_name":"Brown University","ror":"https://ror.org/05gq02987","country_code":"US","type":"education","lineage":["https://openalex.org/I27804330"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Ding, Yiran","raw_affiliation_strings":["Brown University Providence , Rhode Island , USA"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Brown University Providence , Rhode Island , USA","institution_ids":["https://openalex.org/I27804330"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5139742899","display_name":"Yanfu Zhang","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I16285277","display_name":"William & Mary","ror":"https://ror.org/03hsf0573","country_code":"US","type":"education","lineage":["https://openalex.org/I16285277"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Zhang, Yanfu","raw_affiliation_strings":["William & Mary Williamsburg , Virginia , USA"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"William & Mary Williamsburg , Virginia , USA","institution_ids":["https://openalex.org/I16285277"]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":3,"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":0.9771000146865845,"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":0.9771000146865845,"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":0.020899999886751175,"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":0.00019999999494757503,"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/heterogeneous-graph-neural-networks","display_name":"heterogeneous graph neural networks","score":0.9760000109672546},{"id":"https://openalex.org/keywords/explanation-faithfulness","display_name":"explanation fidelity","score":0.5339999794960022},{"id":"https://openalex.org/keywords/shapley-value","display_name":"Shapley value","score":0.6990000009536743},{"id":"https://openalex.org/keywords/han","display_name":"Han","score":0.597000002861023}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.614799976348877},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5152000188827515},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.4339999854564667},{"id":"https://openalex.org/C184337299","wikidata":"https://www.wikidata.org/wiki/Q1437428","display_name":"Semantics (computer science)","level":2,"score":0.4259999990463257},{"id":"https://openalex.org/C132525143","wikidata":"https://www.wikidata.org/wiki/Q141488","display_name":"Graph","level":2,"score":0.3962000012397766},{"id":"https://openalex.org/C107673813","wikidata":"https://www.wikidata.org/wiki/Q812534","display_name":"Bayesian probability","level":2,"score":0.3856000006198883},{"id":"https://openalex.org/C66882249","wikidata":"https://www.wikidata.org/wiki/Q169336","display_name":"Homogeneous","level":2,"score":0.36640000343322754},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.36489999294281006},{"id":"https://openalex.org/C117220453","wikidata":"https://www.wikidata.org/wiki/Q5172842","display_name":"Correlation","level":2,"score":0.3443000018596649},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.3149000108242035},{"id":"https://openalex.org/C4679612","wikidata":"https://www.wikidata.org/wiki/Q866298","display_name":"Aggregate (composite)","level":2,"score":0.3084999918937683},{"id":"https://openalex.org/C185798385","wikidata":"https://www.wikidata.org/wiki/Q1161707","display_name":"Benchmark (surveying)","level":2,"score":0.29179999232292175},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.290800005197525},{"id":"https://openalex.org/C124304363","wikidata":"https://www.wikidata.org/wiki/Q673661","display_name":"Abstraction","level":2,"score":0.28209999203681946},{"id":"https://openalex.org/C108154423","wikidata":"https://www.wikidata.org/wiki/Q1469792","display_name":"Salience (neuroscience)","level":2,"score":0.2711000144481659},{"id":"https://openalex.org/C164226766","wikidata":"https://www.wikidata.org/wiki/Q7293202","display_name":"Rank (graph theory)","level":2,"score":0.26739999651908875},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.2637999951839447},{"id":"https://openalex.org/C36503486","wikidata":"https://www.wikidata.org/wiki/Q11235244","display_name":"Domain (mathematical analysis)","level":2,"score":0.26080000400543213},{"id":"https://openalex.org/C189950617","wikidata":"https://www.wikidata.org/wiki/Q937228","display_name":"Property (philosophy)","level":2,"score":0.2578999996185303},{"id":"https://openalex.org/C121375916","wikidata":"https://www.wikidata.org/wiki/Q936559","display_name":"Principle of compositionality","level":2,"score":0.2547999918460846},{"id":"https://openalex.org/C171018156","wikidata":"https://www.wikidata.org/wiki/Q7370306","display_name":"Rotation formalisms in three dimensions","level":2,"score":0.25380000472068787}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2602.08500","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2602.08500","pdf_url":"https://arxiv.org/pdf/2602.08500","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.2602.08500","doi":"https://doi.org/10.48550/arxiv.2602.08500","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2602.08500","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":"pmh:oai:arXiv.org:2602.08500","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2602.08500","pdf_url":"https://arxiv.org/pdf/2602.08500","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"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7128508703.pdf","grobid_xml":"https://content.openalex.org/works/W7128508703.grobid-xml"},"referenced_works_count":4,"referenced_works":["https://openalex.org/W2169393322","https://openalex.org/W2979481854","https://openalex.org/W4382461866","https://openalex.org/W4406447978"],"related_works":["https://openalex.org/W3210073855","https://openalex.org/W4382461866","https://openalex.org/W7118023658","https://openalex.org/W4389945727","https://openalex.org/W4380632725","https://openalex.org/W3135488505","https://openalex.org/W7127650098","https://openalex.org/W4412825066","https://openalex.org/W4410990384","https://openalex.org/W3173429910"],"abstract_inverted_index":{"Meta-path-based":[0],"heterogeneous":[1,63],"graph":[2],"neural":[3],"networks":[4],"aggregate":[5],"over":[6,13],"meta-path-induced":[7],"views,":[8],"and":[9,40,59,68,103,117,123,127,132,158,181,190,199],"their":[10],"semantic-level":[11],"attention":[12,36,142,156],"meta-path":[14,35,38,92,160],"channels":[15],"is":[16,48],"widely":[17],"used":[18],"as":[19],"a":[20,73,80],"narrative":[21],"for":[22,56],"``which":[23],"semantics":[24,67],"matter.''":[25],"We":[26,112],"study":[27],"this":[28],"assumption":[29],"empirically":[30],"by":[31],"asking:":[32],"when":[33,41],"does":[34],"reflect":[37],"importance,":[39],"can":[42,65],"it":[43],"decouple?":[44],"A":[45],"key":[46],"challenge":[47],"that":[49,85,169],"most":[50],"post-hoc":[51,82],"GNN":[52],"explainers":[53,88,119],"are":[54],"designed":[55],"homogeneous":[57],"graphs,":[58],"naive":[60],"adaptations":[61],"to":[62],"neighborhoods":[64],"mix":[66],"confound":[69],"perturbations.":[70],"To":[71,140],"enable":[72],"controlled":[74],"empirical":[75],"analysis,":[76],"we":[77,144],"introduce":[78],"MetaXplain,":[79],"meta-path-aware":[81,170],"explanation":[83],"protocol":[84],"applies":[86],"existing":[87],"in":[89,200],"the":[90,109,188],"native":[91],"view":[93],"domain":[94],"via":[95],"(i)":[96],"view-factorized":[97],"explanations,":[98],"(ii)":[99],"schema-valid":[100],"channel-wise":[101],"perturbations,":[102],"(iii)":[104],"fusion-aware":[105],"attribution,":[106],"without":[107],"modifying":[108],"underlying":[110],"predictor.":[111],"benchmark":[113],"representative":[114],"gradient-,":[115],"perturbation-,":[116],"Shapley-style":[118],"on":[120,187,194],"ACM,":[121],"DBLP,":[122],"IMDB":[124],"with":[125],"HAN":[126],"HAN-GCN,":[128],"comparing":[129],"against":[130],"xPath":[131],"type-matched":[133],"random":[134,164,174],"baselines":[135],"under":[136],"standard":[137],"faithfulness":[138],"metrics.":[139],"quantify":[141],"reliability,":[143],"propose":[145],"Meta-Path":[146],"Attention--Explanation":[147],"Alignment":[148],"(MP-AEA),":[149],"which":[150],"measures":[151],"rank":[152],"correlation":[153],"between":[154],"learned":[155],"weights":[157],"explanation-derived":[159],"contribution":[161],"scores":[162],"across":[163],"runs.":[165],"Our":[166],"results":[167],"show":[168],"explanations":[171],"typically":[172],"outperform":[173],"controls,":[175],"while":[176],"MP-AEA":[177],"reveals":[178],"both":[179],"high-alignment":[180],"statistically":[182],"significant":[183],"decoupling":[184],"regimes":[185,203],"depending":[186],"dataset":[189],"backbone;":[191],"moreover,":[192],"retraining":[193],"explanation-induced":[195],"subgraphs":[196],"often":[197],"preserves,":[198],"some":[201],"noisy":[202],"improves,":[204],"predictive":[205],"performance,":[206],"suggesting":[207],"an":[208],"explanation-as-denoising":[209],"effect.":[210]},"counts_by_year":[],"updated_date":"2026-11-07T11:25:07.637180","created_date":"2026-02-11T00:00:00"}