CrossHGAT: Cross-Network Heterogeneous Graph Attention Networks for Financial Fraud Detection in Corporate Supply Chains and Investment Networks

CrossHGAT: Cross-Network Heterogeneous Graph Attention Networks for Financial Fraud Detection in Corporate Supply Chains and Investment Networks | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article CrossHGAT: Cross-Network Heterogeneous Graph Attention Networks for Financial Fraud Detection in Corporate Supply Chains and Investment Networks Ning Hu, Jiaxiang Xiao, Yan Shi This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9252427/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Financial fraud in corporate supply chains and investment networks has become increasingly sophisticated, exploiting the entanglement of transaction flows with equity ownership structures in ways that single-domain detection systems fundamentally cannot capture. We propose CrossHGAT, a cross-network heterogeneous graph attention framework that jointly models three semantically distinct graphs---an enterprise graph, a transaction graph, and a novel investment graph---and introduces explicit cross-domain reasoning through a Cross-Graph Bridging Attention Module (CGBAM). Within each graph, a hierarchical encoder performs node-level attention over typed metapath neighbors and semantic-level attention across fraud-discriminative metapaths, producing rich domain-specific embeddings. CGBAM then propagates fraud signals bidirectionally between the supply chain and investment domains via shared entity anchors and a learnable domain-balance gate $\gamma$, enabling detection of camouflaged fraud patterns that span both domains. A three-task multi-task objective with homoscedastic uncertainty weighting jointly optimizes fraud classification for enterprises, transactions, and investment entities, while a multi-view explainer generates cross-graph attribution scores to support interpretable risk decisions. Extensive experiments on four synthetic datasets and two real-world datasets demonstrate that CrossHGAT consistently outperforms state-of-the-art baselines, achieving up to 6.4\% AUC improvement on enterprise fraud detection and 5.1\% on transaction fraud detection over the strongest competitor. Ablation studies confirm the individual contribution of each proposed component, and case studies illustrate CrossHGAT's unique capacity to surface fraud concealed behind legitimate supply chain behavior through investment-side structural reasoning. Financial Fraud Detection Heterogeneous Graph Neural Networks Cross-Graph Attention Multi-task Learning Graph Explainability Full Text Additional Declarations No competing interests reported. Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-9252427","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":629571529,"identity":"4712626f-41ee-4563-b55d-9d572d59b079","order_by":0,"name":"Ning Hu","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAxElEQVRIiWNgGAWjYBACAxDx8Y+NHD8z88EHRGthnNmQZizZzpZsQLQWZs6Gw4kG53nMBIjSYs5+9uBnxh3MCcaHGcwYGGpsoglqsezJS5YuPMOWZ3aYIe0Bw7G03AaCDjuQYyA9g42nGKjluAFjw2EitJx/Y/ybh00icXMzY5sEcVpu5JhJ87YZJG5gZmYjVssbM8sZZxKMJQ6zMRskEOWX8znGNz5U/Jfj7z//8cGHGhvCWlBBAmnKR8EoGAWjYBTgAgCRID/hwb2UXQAAAABJRU5ErkJggg==","orcid":"","institution":"Xinyang College of Agriculture and Forestry","correspondingAuthor":true,"prefix":"","firstName":"Ning","middleName":"","lastName":"Hu","suffix":""},{"id":629571530,"identity":"bb6156ee-897f-4bbd-a846-103ef82e904a","order_by":1,"name":"Jiaxiang Xiao","email":"","orcid":"","institution":"Xinyang College of Agriculture and Forestry","correspondingAuthor":false,"prefix":"","firstName":"Jiaxiang","middleName":"","lastName":"Xiao","suffix":""},{"id":629571531,"identity":"d290ba9c-881c-47eb-baa8-d7bb71441400","order_by":2,"name":"Yan Shi","email":"","orcid":"","institution":"Xinyang College of Agriculture and Forestry","correspondingAuthor":false,"prefix":"","firstName":"Yan","middleName":"","lastName":"Shi","suffix":""}],"badges":[],"createdAt":"2026-03-28 11:53:32","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-9252427/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9252427/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":108355537,"identity":"941dbab1-7240-4286-9f12-611be4ba3128","added_by":"auto","created_at":"2026-05-03 13:10:50","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":4343596,"visible":true,"origin":"","legend":"","description":"","filename":"HeterogeneousGraphAttentionNetworksforFinancialFraudDetectioninComplexCorporateSupplyChainsandInvestmentNetworks.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9252427/v1_covered_6c130667-f450-47cc-8493-edb62103fbb2.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"CrossHGAT: Cross-Network Heterogeneous Graph Attention Networks for Financial Fraud Detection in Corporate Supply Chains and Investment Networks","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":true,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":true,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Financial Fraud Detection, Heterogeneous Graph Neural Networks, Cross-Graph Attention, Multi-task Learning, Graph Explainability","lastPublishedDoi":"10.21203/rs.3.rs-9252427/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9252427/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eFinancial fraud in corporate supply chains and investment networks has become increasingly sophisticated, exploiting the entanglement of transaction flows with equity ownership structures in ways that single-domain detection systems fundamentally cannot capture. We propose CrossHGAT, a cross-network heterogeneous graph attention framework that jointly models three semantically distinct graphs---an enterprise graph, a transaction graph, and a novel investment graph---and introduces explicit cross-domain reasoning through a Cross-Graph Bridging Attention Module (CGBAM). Within each graph, a hierarchical encoder performs node-level attention over typed metapath neighbors and semantic-level attention across fraud-discriminative metapaths, producing rich domain-specific embeddings. CGBAM then propagates fraud signals bidirectionally between the supply chain and investment domains via shared entity anchors and a learnable domain-balance gate $\\gamma$, enabling detection of camouflaged fraud patterns that span both domains. A three-task multi-task objective with homoscedastic uncertainty weighting jointly optimizes fraud classification for enterprises, transactions, and investment entities, while a multi-view explainer generates cross-graph attribution scores to support interpretable risk decisions. Extensive experiments on four synthetic datasets and two real-world datasets demonstrate that CrossHGAT consistently outperforms state-of-the-art baselines, achieving up to 6.4\\% AUC improvement on enterprise fraud detection and 5.1\\% on transaction fraud detection over the strongest competitor. Ablation studies confirm the individual contribution of each proposed component, and case studies illustrate CrossHGAT's unique capacity to surface fraud concealed behind legitimate supply chain behavior through investment-side structural reasoning.\u003c/p\u003e","manuscriptTitle":"CrossHGAT: Cross-Network Heterogeneous Graph Attention Networks for Financial Fraud Detection in Corporate Supply Chains and Investment Networks","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-04-28 12:21:27","doi":"10.21203/rs.3.rs-9252427/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"b44bd072-4e16-4b42-affb-485f51997b2a","owner":[],"postedDate":"April 28th, 2026","published":true,"recentEditorialEvents":[{"type":"decision","content":"Rejected","date":"2026-05-03T12:57:38+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-05-03T09:29:19+00:00","index":28,"fulltext":""}],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2026-05-03T13:10:15+00:00","versionOfRecord":[],"versionCreatedAt":"2026-04-28 12:21:27","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9252427","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9252427","identity":"rs-9252427","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}