C-VARC: A Large-Scale Chinese Value Rule Corpus for Value Alignment of Large Language Models

C-VARC: A Large-Scale Chinese Value Rule Corpus for Value Alignment of Large Language Models | 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 data-descriptor C-VARC: A Large-Scale Chinese Value Rule Corpus for Value Alignment of Large Language Models Ping Wu, Guobin Shen, Dongcheng Zhao, Yuwei Wang, Yiting Dong, and 4 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7901567/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 Ensuring that Large Language Models (LLMs) align with mainstream human values and ethical norms is crucial for the safe and sustainable development of AI. Current value evaluation and alignment are constrained by Western cultural bias and incomplete domestic frameworks reliant on non-native rules; furthermore, the lack of scalable, rule-driven scenario generation methods makes evaluations costly and inadequate across diverse cultural contexts. To address these challenges, we propose a hierarchical value framework grounded in core Chinese values, encompassing three main dimensions, 12 core values, and 50 derived values. Based on this framework, we construct a large-scale Chinese Value Rule Corpus (C-VARC) containing over 250,000 value rules enhanced and expanded through human annotation. Experimental results demonstrate that scenarios guided by C-VARC exhibit clearer value boundaries and greater content diversity compared to those produced through direct generation. In the evaluation across six sensitive themes (e.g., sur-rogacy, suicide), seven mainstream LLMs preferred C-VARC generated options in over 70.5% of cases, while five Chinese human annotators showed an 87.5% alignment with C-VARC, confirming its universality, cultural relevance, and strong alignment with Chinese values. Additionally, we construct 400,000 rule-based moral dilemma scenarios that objectively capture nuanced distinctions in conflicting value prioritization across 17 LLMs. Our work establishes a culturally-adaptive benchmarking framework for comprehensive value evaluation and alignment, representing Chinese characteristics. Full Text Additional Declarations No competing interests reported. Supplementary Files sdatasensitivedatachecklist.docx 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-7901567","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"data-descriptor","associatedPublications":[],"authors":[{"id":558948379,"identity":"8c6d9ed9-8a6d-4a18-8b82-6a9124f9f60f","order_by":0,"name":"Ping Wu","email":"","orcid":"","institution":"BrainCog Lab, Institute of Automation, Chinese Academy of Sciences","correspondingAuthor":false,"prefix":"","firstName":"Ping","middleName":"","lastName":"Wu","suffix":""},{"id":558948389,"identity":"6fdcb76a-2e94-4bdf-84ea-4d353119b1e5","order_by":1,"name":"Guobin Shen","email":"","orcid":"","institution":"BrainCog Lab, Institute of Automation, Chinese Academy of Sciences","correspondingAuthor":false,"prefix":"","firstName":"Guobin","middleName":"","lastName":"Shen","suffix":""},{"id":558948394,"identity":"a0d0959c-a818-4a5a-bebc-d9fde5027524","order_by":2,"name":"Dongcheng Zhao","email":"","orcid":"","institution":"Beijing Key Laboratory of Safe AI and Superalignment","correspondingAuthor":false,"prefix":"","firstName":"Dongcheng","middleName":"","lastName":"Zhao","suffix":""},{"id":558948395,"identity":"2d7363ee-7e35-4d0f-863a-3ac2604072fe","order_by":3,"name":"Yuwei Wang","email":"","orcid":"","institution":"Long-term AI","correspondingAuthor":false,"prefix":"","firstName":"Yuwei","middleName":"","lastName":"Wang","suffix":""},{"id":558948398,"identity":"972ab02a-86ee-4841-8541-6c9f4d0161f8","order_by":4,"name":"Yiting Dong","email":"","orcid":"","institution":"BrainCog Lab, Institute of Automation, Chinese Academy of Sciences","correspondingAuthor":false,"prefix":"","firstName":"Yiting","middleName":"","lastName":"Dong","suffix":""},{"id":558948399,"identity":"fe3ae441-718d-42ec-91f3-c69d9948d302","order_by":5,"name":"Yu Shi","email":"","orcid":"","institution":"The School of Artificial Intelligence, University of Chinese Academy of Sciences","correspondingAuthor":false,"prefix":"","firstName":"Yu","middleName":"","lastName":"Shi","suffix":""},{"id":558948400,"identity":"cd4a1690-4874-49d2-985d-4ccd13490f96","order_by":6,"name":"Enmeng Lu","email":"","orcid":"","institution":"Beijing Key Laboratory of Safe AI and Superalignment","correspondingAuthor":false,"prefix":"","firstName":"Enmeng","middleName":"","lastName":"Lu","suffix":""},{"id":558948401,"identity":"073de973-147f-4811-a51b-89782a762818","order_by":7,"name":"Feifei Zhao","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA/klEQVRIie3QsUrDUBTG8RMupMuBrF+wmFe4JRAcin2VG4R2ydalQyiBQl0EV0Ffom9wQiAuwT5AHW4X57gUp2I3B0kbN4f7gzP+OYdD5Dj/EU6zIASklIjhMfdLGkJYKD+1djjtnVBYEMcjO64uF9Hzam9lcYNgQAlMth0Ghdq3lL93Jt5LHWtpgHBFU5hmxxA/BtUfnYmCSa4+11jqimqkDzsmoYS8ovtCH7MDyiMwqbw10uMbRzI4nE0YWYKyALRSShsW1sLntwDZXEsNoPI9a/iORxXPYeruJHqabazkSwSP27b84tvJ9ev9pm3zHt/+oU5j/hI4juM4v3wDG25QSov1sVgAAAAASUVORK5CYII=","orcid":"","institution":"BrainCog Lab, Institute of Automation, Chinese Academy of Sciences","correspondingAuthor":true,"prefix":"","firstName":"Feifei","middleName":"","lastName":"Zhao","suffix":""},{"id":558948412,"identity":"b7091071-6bec-4c7f-835e-b3c9e098721e","order_by":8,"name":"Yi Zeng","email":"","orcid":"","institution":"BrainCog Lab, Institute of Automation, Chinese Academy of Sciences","correspondingAuthor":false,"prefix":"","firstName":"Yi","middleName":"","lastName":"Zeng","suffix":""}],"badges":[],"createdAt":"2025-10-20 02:53:20","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7901567/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7901567/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":98009397,"identity":"90088aac-9f5f-4180-9e92-4756f7c3039b","added_by":"auto","created_at":"2025-12-11 18:00:51","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":5816483,"visible":true,"origin":"","legend":"","description":"","filename":"CVARCALargeScaleChineseValueAndRuleCorpusforValueAlignmentofLargeLanguageModels.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7901567/v1/2ac36344a77104d088c6d18d.pdf"},{"id":98009394,"identity":"d3b94023-0014-44b7-b018-f4906948cf94","added_by":"auto","created_at":"2025-12-11 18:00:51","extension":"json","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":10525,"visible":true,"origin":"","legend":"","description":"","filename":"a9e7596ea28f46b7b830d8ff417477a2.json","url":"https://assets-eu.researchsquare.com/files/rs-7901567/v1/e769f45f20ffc57ece988e1e.json"},{"id":98009398,"identity":"1a4921f1-eb45-4493-94bb-82e1a390f03d","added_by":"auto","created_at":"2025-12-11 18:00:51","extension":"zip","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":5855343,"visible":true,"origin":"","legend":"","description":"","filename":"CVARCALargeScaleChineseValueAndRuleCorpusforValueAlignmentofLargeLanguageModels.zip","url":"https://assets-eu.researchsquare.com/files/rs-7901567/v1/413da022a9d2bf6a90596597.zip"},{"id":98425174,"identity":"bc25966e-ff52-4ba9-bd03-5ca7f3871940","added_by":"auto","created_at":"2025-12-17 16:34:28","extension":"docx","order_by":3,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":59661,"visible":true,"origin":"","legend":"","description":"","filename":"sdatasensitivedatachecklist.docx","url":"https://assets-eu.researchsquare.com/files/rs-7901567/v1/83f7d7c8e4cbb17b59d74e0b.docx"},{"id":98444188,"identity":"2ecfe42b-3626-4c27-95ae-30e06102ae78","added_by":"auto","created_at":"2025-12-17 17:15:17","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":6317484,"visible":true,"origin":"","legend":"","description":"","filename":"CVARCALargeScaleChineseValueAndRuleCorpusforValueAlignmentofLargeLanguageModels.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7901567/v1_covered_b924c1da-463b-4b0a-8eb8-a9de893aec71.pdf"},{"id":98009395,"identity":"7d645865-ddde-4dfd-b3ca-b9e73d36743d","added_by":"auto","created_at":"2025-12-11 18:00:51","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":59661,"visible":true,"origin":"","legend":"","description":"","filename":"sdatasensitivedatachecklist.docx","url":"https://assets-eu.researchsquare.com/files/rs-7901567/v1/2a91f1c0e6f236686dd3928b.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"C-VARC: A Large-Scale Chinese Value Rule Corpus for Value Alignment of Large Language Models","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":"","lastPublishedDoi":"10.21203/rs.3.rs-7901567/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7901567/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"Ensuring that Large Language Models (LLMs) align with mainstream human values and ethical norms is crucial for the safe and sustainable development of AI. Current value evaluation and alignment are constrained by Western cultural bias and incomplete domestic frameworks reliant on non-native rules; furthermore, the lack of scalable, rule-driven scenario generation methods makes evaluations costly and inadequate across diverse cultural contexts. To address these challenges, we propose a hierarchical value framework grounded in core Chinese values, encompassing three main dimensions, 12 core values, and 50 derived values. Based on this framework, we construct a large-scale Chinese Value Rule Corpus (C-VARC) containing over 250,000 value rules enhanced and expanded through human annotation. Experimental results demonstrate that scenarios guided by C-VARC exhibit clearer value boundaries and greater content diversity compared to those produced through direct generation. In the evaluation across six sensitive themes (e.g., sur-rogacy, suicide), seven mainstream LLMs preferred C-VARC generated options in over 70.5% of cases, while five Chinese human annotators showed an 87.5% alignment with C-VARC, confirming its universality, cultural relevance, and strong alignment with Chinese values. Additionally, we construct 400,000 rule-based moral dilemma scenarios that objectively capture nuanced distinctions in conflicting value prioritization across 17 LLMs. Our work establishes a culturally-adaptive benchmarking framework for comprehensive value evaluation and alignment, representing Chinese characteristics.","manuscriptTitle":"C-VARC: A Large-Scale Chinese Value Rule Corpus for Value Alignment of Large Language Models","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-12-11 18:00:47","doi":"10.21203/rs.3.rs-7901567/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":"3d169bc1-369e-49c6-87bc-816577befe8c","owner":[],"postedDate":"December 11th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2025-12-11T18:00:47+00:00","versionOfRecord":[],"versionCreatedAt":"2025-12-11 18:00:47","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7901567","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7901567","identity":"rs-7901567","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}