T-TPC: A Tri-View Target-Prototype Contrastive Network for Implicit Hate Detection

T-TPC: A Tri-View Target-Prototype Contrastive Network for Implicit Hate Detection | 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 Article T-TPC: A Tri-View Target-Prototype Contrastive Network for Implicit Hate Detection Hayder Ansaf, Yuan Rao This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8066357/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 5 You are reading this latest preprint version Abstract Implicit hate speech poses a persistent challenge for automated detection systems, largely because its meaning is rarely explicit. Instead, it often relies on indirect phrasing, subtle cues, or socially situated language—elements that easily slip past models trained to flag overt slurs or aggressive tone. Although contrastive learning has recently improved how models capture sentence-level semantics, most existing methods still view the input in isolation. This single-perspective framing tends to overlook the deeper interplay between who is being targeted, how they are framed, and what tone or implication underlies the message. To address this, we propose the T-TPC: A Tri-View Target-Prototype Contrastive Network, a model designed to reflect how humans navigate such ambiguity. T-TPC integrates three complementary views: it identifies hate speech targets and models their relationship to surrounding context; it compares each instance to class-level semantic prototypes; and it uses a target-aware hard negative sampling strategy to refine the learning process. Together, these components produce a richer sentence representation that better separates implicit hate from benign content. Experiments on three benchmark datasets—IHC, SBIC, and DYNA—demonstrate that T-TPC achieves new state-of-the-art performance, demonstrating the effectiveness of our integrated, multi-view approach. Physical sciences/Mathematics and computing Biological sciences/Psychology Social science/Psychology Full Text Additional Declarations No competing interests reported. Cite Share Download PDF Status: Under Review Version 1 posted Reviewers invited by journal 28 Apr, 2026 Editor invited by journal 12 Nov, 2025 Editor assigned by journal 12 Nov, 2025 Submission checks completed at journal 12 Nov, 2025 First submitted to journal 08 Nov, 2025 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-8066357","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":631114807,"identity":"cb877b33-8424-45fb-8e97-54c552941218","order_by":0,"name":"Hayder Ansaf","email":"data:image/png;base64,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","orcid":"","institution":"Xi'an Jiaotong University","correspondingAuthor":true,"prefix":"","firstName":"Hayder","middleName":"","lastName":"Ansaf","suffix":""},{"id":631114808,"identity":"ebdf98e8-c24e-4d3a-a3ed-9c54db426df4","order_by":1,"name":"Yuan Rao","email":"","orcid":"","institution":"Xi'an Jiaotong University","correspondingAuthor":false,"prefix":"","firstName":"Yuan","middleName":"","lastName":"Rao","suffix":""}],"badges":[],"createdAt":"2025-11-08 23:53:13","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8066357/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8066357/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":108968241,"identity":"0a4716bd-4331-4048-ad33-3edb002e8d8f","added_by":"auto","created_at":"2026-05-11 10:00:03","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":727643,"visible":true,"origin":"","legend":"","description":"","filename":"TTPCATriViewTargetPrototypeContrastiveNetwork.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8066357/v1_covered_9f39aba6-eb56-4bde-8c0b-d9e99ff0f2dd.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"T-TPC: A Tri-View Target-Prototype Contrastive Network for Implicit Hate Detection","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"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":"scientific-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scirep","sideBox":"Learn more about [Scientific Reports](http://www.nature.com/srep/)","snPcode":"","submissionUrl":"","title":"Scientific Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Scientific Reports","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-8066357/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8066357/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"Implicit hate speech poses a persistent challenge for automated detection systems, largely because its meaning is rarely explicit. Instead, it often relies on indirect phrasing, subtle cues, or socially situated language—elements that easily slip past models trained to flag overt slurs or aggressive tone. Although contrastive learning has recently improved how models capture sentence-level semantics, most existing methods still view the input in isolation. This single-perspective framing tends to overlook the deeper interplay between who is being targeted, how they are framed, and what tone or implication underlies the message. To address this, we propose the T-TPC: A Tri-View Target-Prototype Contrastive Network, a model designed to reflect how humans navigate such ambiguity. T-TPC integrates three complementary views: it identifies hate speech targets and models their relationship to surrounding context; it compares each instance to class-level semantic prototypes; and it uses a target-aware hard negative sampling strategy to refine the learning process. Together, these components produce a richer sentence representation that better separates implicit hate from benign content. Experiments on three benchmark datasets—IHC, SBIC, and DYNA—demonstrate that T-TPC achieves new state-of-the-art performance, demonstrating the effectiveness of our integrated, multi-view approach.","manuscriptTitle":"T-TPC: A Tri-View Target-Prototype Contrastive Network for Implicit Hate Detection","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-05-11 09:56:41","doi":"10.21203/rs.3.rs-8066357/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"reviewersInvited","content":"","date":"2026-04-28T10:35:35+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2025-11-12T20:58:02+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-11-12T05:14:18+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-11-12T05:12:52+00:00","index":"","fulltext":""},{"type":"submitted","content":"Scientific Reports","date":"2025-11-08T23:50:55+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"scientific-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scirep","sideBox":"Learn more about [Scientific Reports](http://www.nature.com/srep/)","snPcode":"","submissionUrl":"","title":"Scientific Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Scientific Reports","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"4bbacb52-580c-4cae-aacc-3c4b0d42cbb4","owner":[],"postedDate":"May 11th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[{"id":67173340,"name":"Physical sciences/Mathematics and computing"},{"id":67173341,"name":"Biological sciences/Psychology"},{"id":67173342,"name":"Social science/Psychology"}],"tags":[],"updatedAt":"2026-05-11T09:56:41+00:00","versionOfRecord":[],"versionCreatedAt":"2026-05-11 09:56:41","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8066357","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8066357","identity":"rs-8066357","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}