Explainable Retrieval-Augmented Generation Framework for Evidence-Aligned and Faithful Legal Reasoning

Explainable Retrieval-Augmented Generation Framework for Evidence-Aligned and Faithful Legal Reasoning | 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 Explainable Retrieval-Augmented Generation Framework for Evidence-Aligned and Faithful Legal Reasoning Nitin Wasudeorao Wankhade This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8947587/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 Retrieval-augmented generation (RAG) has emerged as an effective paradigm for legal question answering, but existing systems often suffer from hallucinated citations and weak evidence grounding. This paper proposes EL-RAG, an explainable retrieval-augmented generation framework designed to ensure evidence-aligned and faithful legal reasoning. The framework formulates evidence grounding as a probabilistic multi-objective optimization problem, integrating a hybrid sparse–dense retriever, a learning-to-rank reranker, a multi-hop reasoning generator, and an Evidence Alignment Layer (EAL). Two evaluation metrics—Citation Alignment Score (CAS) and Faithful Justification Index (FJI) are introduced to assess grounding and interpretability. Experiments on COLIEE 2025, NyayaRAG 2025, CaseLawBench, and LegalBench-RAG show consistent improvements over strong baselines, significantly reducing hallucinations and improving expert-rated interpretability. Artificial Intelligence and Machine Learning Retrieval-Augmented Generation Explainable Artificial Intelligence Legal Reasoning Evidence Alignment Full Text Additional Declarations The authors declare no competing interests. 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-8947587","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":595680537,"identity":"b64c0cd7-8793-4355-8935-06fa1daeffc8","order_by":0,"name":"Nitin Wasudeorao Wankhade","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA9klEQVRIiWNgGAWjYDACZgaGA2AGDw8QM9gAWYyNB/BrYUbRkgbS0oBfC8gaJC2HwWy8WszZ+Q8eLmA4LG/Oc/bghzcV5+3Wth8G2lJjE41Li2UzM8PhGQyHDXf29iVLzjlzO3nbmUSglmNpuQ04tBgcBmrhYbjNuOE8j4E0b9vtZLMDQC2MDYcJarEHajH+zfvvXLLZ+YfEaUnccLbHTJq34YCd2Q3Cthgc5jH4n7zhzBkzyznHkhPMbgBtScDnl/MHH3/mqUiz3XAmx/jGmxo7e7Pz6Q8ffKixwakFqhHBTASrTMCrHA3Yk6J4FIyCUTAKRgYAAPQhY5MnFmcNAAAAAElFTkSuQmCC","orcid":"","institution":"Nutan Maharashtra Institute of Engineering and Technology (NMIET), Pune, India","correspondingAuthor":true,"prefix":"","firstName":"Nitin","middleName":"Wasudeorao","lastName":"Wankhade","suffix":""}],"badges":[],"createdAt":"2026-02-23 13:24:02","currentVersionCode":1,"declarations":{"humanSubjects":false,"vertebrateSubjects":true,"conflictsOfInterestStatement":false,"humanSubjectEthicalGuidelines":false,"humanSubjectConsent":false,"humanSubjectClinicalTrial":false,"humanSubjectCaseReport":false,"vertebrateSubjectEthicalGuidelines":true},"doi":"10.21203/rs.3.rs-8947587/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8947587/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":103507101,"identity":"254d4da7-658f-4b17-8ef8-e5aa7cf17630","added_by":"auto","created_at":"2026-02-26 13:40:26","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":448910,"visible":true,"origin":"","legend":"","description":"","filename":"Researchpaper.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8947587/v1_covered_7e1c6d50-b34a-4c2e-85aa-42ec1f41d919.pdf"}],"financialInterests":"The authors declare no competing interests.","formattedTitle":"\u003cp\u003eExplainable Retrieval-Augmented Generation Framework for Evidence-Aligned and Faithful Legal Reasoning\u003c/p\u003e","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"Nutan Maharashtra Institute of Engineering and Technology (NMIET), Pune, India","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":"Retrieval-Augmented Generation, Explainable Artificial Intelligence, Legal Reasoning, Evidence Alignment","lastPublishedDoi":"10.21203/rs.3.rs-8947587/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8947587/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eRetrieval-augmented generation (RAG) has emerged as an effective paradigm for legal question answering, but existing systems often suffer from hallucinated citations and weak evidence grounding. This paper proposes EL-RAG, an explainable retrieval-augmented generation framework designed to ensure evidence-aligned and faithful legal reasoning. The framework formulates evidence grounding as a probabilistic multi-objective optimization problem, integrating a hybrid sparse–dense retriever, a learning-to-rank reranker, a multi-hop reasoning generator, and an Evidence Alignment Layer (EAL). Two evaluation metrics—Citation Alignment Score (CAS) and Faithful Justification Index (FJI) are introduced to assess grounding and interpretability. Experiments on COLIEE 2025, NyayaRAG 2025, CaseLawBench, and LegalBench-RAG show consistent improvements over strong baselines, significantly reducing hallucinations and improving expert-rated interpretability.\u003c/p\u003e","manuscriptTitle":"Explainable Retrieval-Augmented Generation Framework for Evidence-Aligned and Faithful Legal Reasoning","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-02-25 00:42:03","doi":"10.21203/rs.3.rs-8947587/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":"22833608-8145-4b41-a854-fd6b6a4e6d7a","owner":[],"postedDate":"February 25th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":63383296,"name":"Artificial Intelligence and Machine Learning"}],"tags":[],"updatedAt":"2026-03-10T06:26:04+00:00","versionOfRecord":[],"versionCreatedAt":"2026-02-25 00:42:03","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8947587","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8947587","identity":"rs-8947587","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}