PlagiarismGuard: Democratizing Academic Integrity with a Free, Open-Source, Multi-API Plagiarism Detection Ecosystem

PlagiarismGuard: Democratizing Academic Integrity with a Free, Open-Source, Multi-API Plagiarism Detection Ecosystem | 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 PlagiarismGuard: Democratizing Academic Integrity with a Free, Open-Source, Multi-API Plagiarism Detection Ecosystem Siddalingaiah H S This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8988484/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 12 You are reading this latest preprint version Abstract Academic integrity is facing an unprecedented crisis in the digital era, exacerbated by the proliferation of "paper mills" and the rapid adoption of sophisticated Generative AI tools. While commercial detection solutions like iThenticate and Turnitin have established themselves as industry standards, their prohibitive cost models create a significant "integrity divide," disproportionately affecting institutions in low- and middle-income countries (LMICs) and independent researchers. This unavailability of affordable verification tools risks compromising the quality of scientific output from resource-constrained regions. Addressing this disparity, we present PlagiarismGuard, a free, open-source, client-side academic integrity verification platform. Built on a modern serverless architecture, PlagiarismGuard aggregates search results from 16 open academic databases—including OpenAlex, Semantic Scholar, and CrossRef—to perform comprehensive similarity analysis without requiring institutional subscriptions. The system implements advanced text fingerprinting algorithms (n-gram shingling), code plagiarism detection (Winnowing), and perceptual image hashing, while also integrating a "Bring Your Own Key" (BYOK) model for AI-powered authorship analysis using models like Google Gemini and OpenAI GPT-4. We detail the system's architecture, privacy-preserving client-side processing model, and performance benchmarks, demonstrating that open-source alternatives can provide robust plagiarism detection capabilities comparable to commercial tools. By releasing PlagiarismGuard under the MIT License, we aim to democratize access to academic integrity tools and foster a community-driven approach to combating research misconduct. plagiarism detection academic integrity educational technology open-source software generative AI digital equity higher education Figures Figure 1 Figure 2 Figure 3 Figure 4 1. Introduction 1.1 The Evolving Landscape of Academic Integrity The preservation of academic integrity is fundamental to the credibility of the scientific enterprise and the educational process (Bretag & Mahmud, 2009 ). Traditionally, plagiarism was defined largely by the unauthorized appropriation of another's text or ideas. However, the digital revolution has significantly transformed this landscape. The immense availability of online information has made "copy-paste" plagiarism effortless, while the recent emergence of Large Language Models (LLMs) has introduced "AI-giarism"—the generation of synthetic text that evades traditional matching algorithms (Weber-Wulff et al., 2023 ). Educational institutions and publishers rely heavily on text-matching software to police these boundaries. Tools like Turnitin and iThenticate have become the gatekeepers of academic publishing, used by thousands of universities and virtually all major publishing houses (Foltýnek et al., 2020 ). These systems rely on massive, proprietary databases of academic content and student submissions to identify overlap. While effective, this centralization of academic policing power in the hands of a few commercial entities has created significant vulnerabilities and inequities in the global research ecosystem. 1.2 The "Integrity Divide" and Economic Barriers A critical, yet often overlooked, consequence of the commercialization of plagiarism detection is the creation of an "integrity divide." High subscription costs, often exceeding tens of thousands of dollars annually for institutional licenses, effectively bar researchers and universities in Low- and Middle-Income Countries (LMICs) from accessing these essential tools (Coughlan, 2019 ). Studies indicate that researchers from resource-constrained settings face higher rejection rates, partly due to unintentional linguistic overlap or formatting issues that could have been resolved with pre-submission screening (Pupovac & Fanelli, 2015 ). Without access to screening tools, these scholars are forced to submit "blind," risking reputational damage and sanctions for unintentional misconduct. 1.3 The Limitations of Proprietary Solutions Beyond cost, proprietary solutions present inherent limitations regarding transparency and data sovereignty. The "black box" nature of commercial algorithms means that educators and researchers cannot verify why a certain phrase was flagged or, conversely, why a paraphrased section was missed. Furthermore, the storage of student and researcher manuscripts in private commercial servers raises significant data privacy and intellectual property concerns (Tennant et al., 2016 ). 1.4 Study Objectives To address these challenges, we developed PlagiarismGuard, a transparent, open-source academic integrity platform. This study describes the design, implementation, and evaluation of PlagiarismGuard. Our primary objectives were: (1) to engineer a zero-cost detection system leveraging global open-access APIs; (2) to implement privacy-preserving client-side processing architecture; (3) to integrate multi-modal detection capabilities (text, code, image); and (4) to provide a "Bring Your Own Key" (BYOK) interface for advanced AI-based authorship analysis. 2. System Design and Architecture 2.1 Architectural Overview PlagiarismGuard is architected as a serverless, client-heavy Progressive Web Application (PWA) to maximize scalability and minimize operational costs (Fig. 1 ). The core application logic resides in the client's browser, developed using React.js v18+. This design choice is deliberate; by performing text preprocessing, tokenization, and report generation on the client side, we eliminate the need for expensive backend computational resources and ensure that the user's full manuscript text is never stored on our servers, addressing the data privacy concerns inherent in cloud-based commercial solutions. 2.2 The Open-API Aggregation Engine Unlike commercial competitors that maintain proprietary databases, PlagiarismGuard acts as a meta-search aggregator. It queries 16 distinct high-quality academic and web databases simultaneously (Table 1 ). These include: Table 1 Complete list of academic and reference databases integrated into PlagiarismGuard through API aggregation. Category Source Specialty API Type Academic Semantic Scholar AI-indexed literature REST Academic OpenAlex Global research graph REST Academic Europe PMC Biomedical sciences REST Academic CrossRef DOI registration REST Academic CORE Open access repository REST Academic arXiv Physics, Math, CS preprints REST Technical IEEE Xplore Engineering standards REST Technical StackExchange Developer Q&A REST Technical GitHub Open source code REST Technical Springer Scientific journals REST Books Google Books Digitized books REST Books Open Library Universal catalog REST Archives Internet Archive Web history REST General DuckDuckGo Web search fallback REST General Wikipedia Reference articles REST AI Google Gemini Authorship detection REST Open Scholarly Citation Indexes: OpenAlex, CrossRef, Semantic Scholar Repository Aggregators: CORE, arXiv, Europe PMC Technical & Code Repositories: GitHub, StackExchange, IEEE Xplore General Web & Book Indices: Google Books, Wikipedia, Internet Archive To manage this high-volume concurrent querying without hitting browser Cross-Origin Resource Sharing (CORS) limits, we implemented a lightweight serverless proxy using Vercel Edge Functions. This proxy handles authentication, request rate-limiting, and response normalization, presenting a unified JSON schema to the frontend client. 2.3 Algorithmic Core We implemented three distinct algorithmic approaches for multi-modal plagiarism detection (Table 2 ). Table 2 Comparison of detection algorithms implemented in PlagiarismGuard with their specific use cases and accuracy characteristics. Algorithm Use Case Accuracy Speed Resilience N-Gram Shingling Text similarity High Fast Word reordering, synonyms TF-IDF Cosine Topical matching Medium Fast Paraphrasing Winnowing Code plagiarism High Medium Variable renaming, whitespace Perceptual Hash Image detection High Fast Resize, compression, format Note: Resilience indicates types of modifications the algorithm can detect through. 2.3.1 Text Similarity: N-Gram Shingling For rapid and robust text similarity detection, we implemented the N-Gram Shingling algorithm (Broder et al., 1997 ). The document D is converted into a set of contiguous subsequences (shingles) of length n (where n = 5 to 9 words). The similarity between a suspect document A and a source document B is calculated using the Jaccard Coefficient: J(A,B) = |S(A) ∩ S(B)| / |S(A) ∪ S(B)| This approach is computationally efficient and resistant to minor rearrangements of words, making it superior to simple substring matching. 2.3.2 Code Plagiarism: Winnowing Recognizing the rise of Computer Science education, PlagiarismGuard includes a dedicated code analysis module. We utilized the Winnowing algorithm (Schleimer et al., 2003 ), a local fingerprinting algorithm utilized by MOSS (Measure of Software Similarity). The algorithm tokenizes source code (ignoring comments and whitespace), generates a stream of hashes, and selects a subset of these hashes (fingerprints) based on a sliding window. This guarantees that any match of text at least as long as the window size is detected, offering robust performance against variable renaming and reformatting attacks. 2.3.3 Visual Similarity: Perceptual Hashing To detect image plagiarism—critical for figures in scientific manuscripts—we implemented a perceptual hashing (pHash) algorithm (Zauner, 2010 ). Unlike cryptographic hashes (like MD5) where a single pixel change results in a completely different hash, perceptual hashes adapt so that similar images produce similar hash values. We use the Discrete Cosine Transform (DCT) to generate a "fingerprint" of the image's low-frequency components, allowing the system to detect images that have been resized, compressed, or slightly color-corrected. 2.4 AI Authorship Analysis PlagiarismGuard adheres to a "neutral platform" philosophy regarding AI detection. Instead of building a proprietary black-box classifier, it integrates with state-of-the-art external Large Language Models (LLMs) via API. Users can bring their own API keys (BYOK) for Google Gemini, OpenAI, or Anthropic Claude. The system constructs specialized prompts designed to analyze linguistic features—such as perplexity, burstiness, and semantic coherence—to estimate the likelihood of AI generation. This approach ensures the tool remains up-to-date with the latest foundational models, unlike static classifiers that quickly become obsolete. 3. Results 3.1 Implementation and Deployment The complete system has been open-sourced on GitHub and deployed as a live service reachable via two mirrors to ensure redundancy: https://plagiarism-checker-web-app.vercel.app/ (Primary) and https://plagiarismguardpro.netlify.app/ (Secondary). The application features a responsive Material Design interface that adapts to desktop, tablet, and mobile workflows. As a PWA, it supports offline installation, allowing users in low-bandwidth environments to load the interface and queue documents for analysis. 3.2 Functional Performance & Reporting The platform successfully processes standard manuscript formats (.txt, .docx, .pdf). The main analysis interface provides real-time feedback and metrics (Fig. 2 ). Upon analysis, users receive a comprehensive "plagiarism report" certificate (Fig. 3 ). This report includes: Similarity Score Validation: A global percentage representing the total overlap with external sources. Source Breakdown: An itemized list of all detected sources, categorized by type (Academic Journal, Website, Repository). Textual Highlighting: Direct side-by-side comparison of the user's text against the matched external source, utilizing color-coded highlights for easy visual auditing. Verification Metadata: A QR code and cryptographic hash of the report, ensuring the certificate's authenticity and preventing tampering. 3.3 Performance Benchmarking To evaluate system efficiency, we benchmarked PlagiarismGuard against a standard 3,000-word academic manuscript. On a standard broadband connection: Initialization Time: < 1.5 seconds to load the core application shell. API Aggregation Latency: The parallel execution of 16 API queries typically resolves within 4–8 seconds. Processing Throughput: The client-side tokenization and shingling engine processes approximately 10,000 words per second on an average consumer laptop (Intel i5 equivalent). This performance profile confirms that client-side JavaScript execution is more than sufficient for real-time document analysis, challenging the assumption that such heavy lifting requires dedicated server clusters. 3.4 Comparison with Commercial Tools Table 3 presents a comparative analysis. While commercial tools maintain larger proprietary databases, PlagiarismGuard offers comparable algorithmic sophistication with significant cost advantages. Note: Database coverage is simulated via aggregate API access. Table 3 Comparative analysis of PlagiarismGuard against leading commercial plagiarism detection tools. Feature PlagiarismGuard Turnitin iThenticate Copyscape Cost Free (MIT License) $ 3-15k/year $ 50k+/year $ 0.05/page Database Size 16 Open APIs 100B+ pages 200M+ articles Web Index AI Detection BYOK (Gemini/GPT) Proprietary Proprietary None Privacy Client-side only Cloud storage Cloud storage Cloud storage Note: Commercial pricing is approximate and varies by institution size and contract terms. 3.5 Mobile Accessibility & PWA Being a Progressive Web Application, PlagiarismGuard can be installed on mobile devices for quick checks (Fig. 4 ). The responsive interface adapts seamlessly from desktop to smartphone, with touch-optimized 44px minimum tap targets. This zero-cost accessibility directly addresses the "integrity divide" faced by LMIC institutions. 3.6 User Privacy A key result of this architectural approach is "Privacy by Design." We verified via network packet inspection that the full manuscript text body is never transmitted to our servers. Only search queries (3–5 word phrases) and statistical metadata are sent to the API proxy. This offers a level of intellectual property protection that is structurally impossible for centralized cloud-based competitors to match. 4. Discussion 4.1 Empowering the "Long Tail" of Academia The primary contribution of PlagiarismGuard is the democratization of academic verification. By removing the cost barrier, we empower the "long tail" of the academic community—independent scholars, students in underfunded institutions, and researchers in developing nations. This aligns with the United Nations Sustainable Development Goal 4 (Quality Education) by ensuring equitable access to technical tools required for academic success. 4.2 The Role of Open Source in EdTech PlagiarismGuard demonstrates the viability of Open Source Software (OSS) in the educational technology sector. Unlike closed proprietary systems, OSS allows institutions to audit the matching algorithms for bias (e.g., checking if the tool adequately indexes non-English repositories). Furthermore, the extensibility of our plugin architecture allows developers to add new data sources (such as regional language databases) without waiting for a commercial vendor to prioritize them. 4.3 Limitations While promising, our approach has limitations. First, reliance on public APIs means coverage is strictly limited to Open Access content; paywalled journals indexed deeply by Turnitin may occasionally be missed, though abstract-level matching often mitigates this. Second, the "BYOK" model for AI detection shifts the cost of API tokens to the user, although this cost is negligible compared to annual subscriptions. Finally, as a client-side tool, extremely large documents (e.g., whole theses > 100MB) may face browser memory constraints. 4.4 Future Directions Future development will focus on three areas: (1) Developing a Learning Management System (LMS) plugin (LTI standard) for seamless integration with Moodle and Canvas; (2) Enhancing cross-language plagiarism detection using vector embeddings; and (3) Creating a federated database network where institutions can opt-in to share anonymized fingerprints, creating a community-sourced comparison database. 5. Conclusions PlagiarismGuard represents a paradigm shift from "plagiarism policing" to "integrity empowerment." By providing a free, transparent, and privacy-respecting tool, we offer a viable alternative to the commercial oligopoly in academic verification. We invite the educational community to adopt, fork, and improve this platform to ensure that financial resources never determine an institution's ability to uphold academic standards. Declarations Funding: No funding was received for this work. Conflicts of Interest: The author declares no conflicts of interest. The software is free and open-source. Availability of Data and Material: Source code is available at GitHub (https://github.com/hssling/Plagiarism_Checker_Web_App) under the MIT License. Live deployment is accessible at https://plagiarism-checker-web-app.vercel.app/ and https://plagiarismguardpro.netlify.app/. Ethics Approval: Not applicable. This study describes software development and does not involve human subjects or personal data. Clinical trial registration: Not applicable. This study describes software development and does not involve human subjects or personal data Hence clinical trial registration number and date of registration are not applicable. Consent to Publish declaration: not applicable Consent to Participate declaration: not applicable References Bretag, T., & Mahmud, S. (2009). Self-plagiarism or appropriate textual re-use? Journal of Academic Ethics, 7(3), 193-205. https://doi.org/10.1007/s10805-009-9092-1 Broder, A. Z., Glassman, S. C., Manasse, M. S., & Zweig, G. (1997). Syntactic clustering of the web. Computer Networks and ISDN Systems, 29(8-13), 1157-1166. https://doi.org/10.1016/S0169-7552(97)00031-7 Coughlan, S. (2019). Universities spend millions on plagiarism detection. BBC News. Retrieved from https://www.bbc.com/news/education-50420455 Foltýnek, T., Meuschke, N., & Gipp, B. (2020). Academic plagiarism detection: A systematic literature review. ACM Computing Surveys, 52(6), 1-42. https://doi.org/10.1145/3345317 Google. (2025). Gemini API Documentation. Retrieved from https://ai.google.dev/docs ICMJE. (2024). Recommendations for the Conduct, Reporting, Editing, and Publication of Scholarly Work in Medical Journals. Retrieved from https://www.icmje.org/recommendations/ Mozilla Laboratory. (2024). Cross-Origin Resource Sharing (CORS). MDN Web Docs. Retrieved from https://developer.mozilla.org/en-US/docs/Web/HTTP/CORS Priem, J., Piwowar, H., & Orr, R. (2022). OpenAlex: A fully-open index of scholarly works, authors, venues, institutions, and concepts. arXiv preprint arXiv:2205.01833 Pupovac, V., & Fanelli, D. (2015). Scientists admitting to plagiarism: A meta-analysis of surveys. Science and Engineering Ethics, 21(5), 1331-1352 Resnik, D. B., & Master, Z. (2013). Policies and initiatives aimed at addressing research misconduct in high-income countries. PLoS Medicine, 10(3), e1001406 Salton, G., & Buckley, C. (1988). Term-weighting approaches in automatic text retrieval. Information Processing & Management, 24(5), 513-523 Schleimer, S., Wilkerson, D. S., & Aiken, A. (2003). Winnowing: Local algorithms for document fingerprinting. Proceedings of the 2003 ACM SIGMOD International Conference on Management of Data, 76-85 Tennant, J. P., Waldner, F., Jacques, D. C., et al. (2016). The academic, economic and societal impacts of Open Access: an evidence-based review. F1000Research, 5, 632 Turnitin. (2025). iThenticate: Professional Plagiarism Prevention. Retrieved from https://www.ithenticate.com/ WAME. (2023). Recommendations on Publication Ethics Policies for Medical Journals. World Association of Medical Editors. Retrieved from https://wame.org/ W3C. (2023). Web Application Manifest. W3C Working Draft. Retrieved from https://www.w3.org/TR/appmanifest/ Weber-Wulff, D. (2014). False Feathers: A Perspective on Academic Plagiarism. Berlin: Springer Weber-Wulff, D., Anohina-Naumeca, A., Bjelobaba, S., et al. (2023). Testing of detection tools for AI-generated text. International Journal for Educational Integrity, 19(1), 26 Zauner, C. (2010). Implementation and benchmarking of perceptual image hash functions [Master's thesis]. Hagenberg, Austria: Upper Austria University of Applied Sciences. Additional Declarations No competing interests reported. Cite Share Download PDF Status: Under Review Version 1 posted Reviews received at journal 09 May, 2026 Reviewers agreed at journal 06 May, 2026 Reviews received at journal 19 Apr, 2026 Reviewers agreed at journal 14 Apr, 2026 Reviews received at journal 09 Apr, 2026 Reviewers agreed at journal 09 Apr, 2026 Reviewers agreed at journal 08 Apr, 2026 Reviewers agreed at journal 06 Apr, 2026 Reviewers invited by journal 06 Apr, 2026 Editor assigned by journal 03 Mar, 2026 Submission checks completed at journal 03 Mar, 2026 First submitted to journal 27 Feb, 2026 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-8988484","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":620294060,"identity":"b4e70a2b-f48e-4894-8a44-693bea2d62d3","order_by":0,"name":"Siddalingaiah H S","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAw0lEQVRIiWNgGAWjYBACAwYGxgMJDAxyDAw8xGthAGkxJlELECc2EK3FnP/4gwMPd9Smbzh+9uCDDwx2croNBLRYzsgxOJB45njuhjN5yYYzGJKNzQ4QctgNHoYDiW3HcjccyDGTBrG3EdRyHugwoJZ0g/NviNVyIAHosLaaBIMbRNtyA+SXtgOGM2+8MTacYUCMX84ff/jwZ1udPN/5HMMHHyrs5AhqgYLDDApglQbEKQeBOgb5BuJVj4JRMApGwQgDAEutS5uyZWbkAAAAAElFTkSuQmCC","orcid":"","institution":"Shridevi Institute of Medical Sciences and Research Hospital","correspondingAuthor":true,"prefix":"","firstName":"Siddalingaiah","middleName":"H","lastName":"S","suffix":""}],"badges":[],"createdAt":"2026-02-27 13:23:27","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8988484/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8988484/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":106636928,"identity":"c2893a38-6bc9-4373-a3d2-a964ee81bd17","added_by":"auto","created_at":"2026-04-10 16:57:34","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":742575,"visible":true,"origin":"","legend":"\u003cp\u003eSystem Architecture\u003c/p\u003e\n\u003cp\u003eSystem architecture diagram showing the serverless proxy pattern connecting the React PWA client to multiple academic APIs. The client-side engine handles document processing and analysis, while the serverless proxy mediates API requests to overcome CORS restrictions and manage rate limiting.\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-8988484/v1/4e34f5b60cad4654a9640639.png"},{"id":106959050,"identity":"2152a6ee-ee26-4050-ae1e-b1565bce1fb2","added_by":"auto","created_at":"2026-04-15 08:44:21","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":1110901,"visible":true,"origin":"","legend":"\u003cp\u003eAnalysis Dashboard Interface\u003c/p\u003e\n\u003cp\u003eScreenshot of the PlagiarismGuard analysis dashboard displaying similarity metrics and detected sources. The interface shows: (A) Overall similarity score with color-coded status indicator, (B) Metrics grid with word count, sources checked, and originality index, (C) Side-by-side document comparison with highlighted matches.\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-8988484/v1/fd00a22cb82049f1c66de980.png"},{"id":106636929,"identity":"3aa4aae5-4bf4-40da-8ec7-68a9856344e9","added_by":"auto","created_at":"2026-04-10 16:57:34","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":1209534,"visible":true,"origin":"","legend":"\u003cp\u003ePDF Certificate Sample\u003c/p\u003e\n\u003cp\u003eSample PDF certificate generated by PlagiarismGuard showing the executive dashboard layout. Features include: (A) Professional header banner with certificate ID, (B) Large circular score indicator with color-coded status badge, (C) Eight-metric grid displaying analysis statistics, (D) QR code containing verification data.\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-8988484/v1/3ca069bef251ca40898d0d08.png"},{"id":106636931,"identity":"699f02d3-87b1-4f89-9b1c-e6ec585b7d63","added_by":"auto","created_at":"2026-04-10 16:57:34","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":1462670,"visible":true,"origin":"","legend":"\u003cp\u003eMobile PWA Interface\u003c/p\u003e\n\u003cp\u003eProgressive Web Application (PWA) installation and mobile interface demonstration. Shows: (A) Browser \"Add to Home Screen\" prompt, (B) Installed app icon on device home screen, (C) Standalone app window without browser UI, (D) Responsive stacked comparison view optimized for portrait mobile screens.\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-8988484/v1/f952999c6016d0279bbb6c4e.png"},{"id":107480442,"identity":"8f9343df-362c-4c13-83a1-86d7602f139a","added_by":"auto","created_at":"2026-04-22 02:10:32","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":5264827,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8988484/v1/37e8c26a-ed63-43f0-a53f-2f03c684f52c.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"PlagiarismGuard: Democratizing Academic Integrity with a Free, Open-Source, Multi-API Plagiarism Detection Ecosystem","fulltext":[{"header":"1. Introduction","content":"\u003cdiv id=\"Sec2\" class=\"Section2\"\u003e \u003ch2\u003e1.1 The Evolving Landscape of Academic Integrity\u003c/h2\u003e \u003cp\u003eThe preservation of academic integrity is fundamental to the credibility of the scientific enterprise and the educational process (Bretag \u0026amp; Mahmud, \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2009\u003c/span\u003e). Traditionally, plagiarism was defined largely by the unauthorized appropriation of another's text or ideas. However, the digital revolution has significantly transformed this landscape. The immense availability of online information has made \"copy-paste\" plagiarism effortless, while the recent emergence of Large Language Models (LLMs) has introduced \"AI-giarism\"\u0026mdash;the generation of synthetic text that evades traditional matching algorithms (Weber-Wulff et al., \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e2023\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eEducational institutions and publishers rely heavily on text-matching software to police these boundaries. Tools like Turnitin and iThenticate have become the gatekeepers of academic publishing, used by thousands of universities and virtually all major publishing houses (Folt\u0026yacute;nek et al., \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e2020\u003c/span\u003e). These systems rely on massive, proprietary databases of academic content and student submissions to identify overlap. While effective, this centralization of academic policing power in the hands of a few commercial entities has created significant vulnerabilities and inequities in the global research ecosystem.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003e1.2 The \"Integrity Divide\" and Economic Barriers\u003c/h2\u003e \u003cp\u003eA critical, yet often overlooked, consequence of the commercialization of plagiarism detection is the creation of an \"integrity divide.\" High subscription costs, often exceeding tens of thousands of dollars annually for institutional licenses, effectively bar researchers and universities in Low- and Middle-Income Countries (LMICs) from accessing these essential tools (Coughlan, \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). Studies indicate that researchers from resource-constrained settings face higher rejection rates, partly due to unintentional linguistic overlap or formatting issues that could have been resolved with pre-submission screening (Pupovac \u0026amp; Fanelli, \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e2015\u003c/span\u003e). Without access to screening tools, these scholars are forced to submit \"blind,\" risking reputational damage and sanctions for unintentional misconduct.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003e1.3 The Limitations of Proprietary Solutions\u003c/h2\u003e \u003cp\u003eBeyond cost, proprietary solutions present inherent limitations regarding transparency and data sovereignty. The \"black box\" nature of commercial algorithms means that educators and researchers cannot verify why a certain phrase was flagged or, conversely, why a paraphrased section was missed. Furthermore, the storage of student and researcher manuscripts in private commercial servers raises significant data privacy and intellectual property concerns (Tennant et al., \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e2016\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003e1.4 Study Objectives\u003c/h2\u003e \u003cp\u003eTo address these challenges, we developed PlagiarismGuard, a transparent, open-source academic integrity platform. This study describes the design, implementation, and evaluation of PlagiarismGuard. Our primary objectives were: (1) to engineer a zero-cost detection system leveraging global open-access APIs; (2) to implement privacy-preserving client-side processing architecture; (3) to integrate multi-modal detection capabilities (text, code, image); and (4) to provide a \"Bring Your Own Key\" (BYOK) interface for advanced AI-based authorship analysis.\u003c/p\u003e \u003c/div\u003e"},{"header":"2. System Design and Architecture","content":"\u003cdiv id=\"Sec7\" class=\"Section2\"\u003e\n \u003ch2\u003e2.1 Architectural Overview\u003c/h2\u003e\n \u003cp\u003ePlagiarismGuard is architected as a serverless, client-heavy Progressive Web Application (PWA) to maximize scalability and minimize operational costs (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e1\u003c/span\u003e). The core application logic resides in the client\u0026apos;s browser, developed using React.js v18+. This design choice is deliberate; by performing text preprocessing, tokenization, and report generation on the client side, we eliminate the need for expensive backend computational resources and ensure that the user\u0026apos;s full manuscript text is never stored on our servers, addressing the data privacy concerns inherent in cloud-based commercial solutions.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec8\" class=\"Section2\"\u003e\n \u003ch2\u003e2.2 The Open-API Aggregation Engine\u003c/h2\u003e\n \u003cp\u003eUnlike commercial competitors that maintain proprietary databases, PlagiarismGuard acts as a meta-search aggregator. It queries 16 distinct high-quality academic and web databases simultaneously (Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). These include:\u003c/p\u003e\n \u003cdiv class=\"gridtable\"\u003e\u0026nbsp;\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e\n \u003ccaption language=\"En\"\u003e\n \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e\n \u003cdiv class=\"CaptionContent\"\u003e\n \u003cp\u003eComplete list of academic and reference databases integrated into PlagiarismGuard through API aggregation.\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003ccolgroup cols=\"4\"\u003e\u003c/colgroup\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\" colname=\"c1\"\u003e\n \u003cp\u003eCategory\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\" colname=\"c2\"\u003e\n \u003cp\u003eSource\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\" colname=\"c3\"\u003e\n \u003cp\u003eSpecialty\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\" colname=\"c4\"\u003e\n \u003cp\u003eAPI Type\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colname=\"c1\"\u003e\n \u003cp\u003eAcademic\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c2\"\u003e\n \u003cp\u003eSemantic Scholar\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c3\"\u003e\n \u003cp\u003eAI-indexed literature\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c4\"\u003e\n \u003cp\u003eREST\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colname=\"c1\"\u003e\n \u003cp\u003eAcademic\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c2\"\u003e\n \u003cp\u003eOpenAlex\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c3\"\u003e\n \u003cp\u003eGlobal research graph\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c4\"\u003e\n \u003cp\u003eREST\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colname=\"c1\"\u003e\n \u003cp\u003eAcademic\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c2\"\u003e\n \u003cp\u003eEurope PMC\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c3\"\u003e\n \u003cp\u003eBiomedical sciences\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c4\"\u003e\n \u003cp\u003eREST\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colname=\"c1\"\u003e\n \u003cp\u003eAcademic\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c2\"\u003e\n \u003cp\u003eCrossRef\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c3\"\u003e\n \u003cp\u003eDOI registration\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c4\"\u003e\n \u003cp\u003eREST\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colname=\"c1\"\u003e\n \u003cp\u003eAcademic\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c2\"\u003e\n \u003cp\u003eCORE\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c3\"\u003e\n \u003cp\u003eOpen access repository\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c4\"\u003e\n \u003cp\u003eREST\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colname=\"c1\"\u003e\n \u003cp\u003eAcademic\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c2\"\u003e\n \u003cp\u003earXiv\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c3\"\u003e\n \u003cp\u003ePhysics, Math, CS preprints\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c4\"\u003e\n \u003cp\u003eREST\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colname=\"c1\"\u003e\n \u003cp\u003eTechnical\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c2\"\u003e\n \u003cp\u003eIEEE Xplore\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c3\"\u003e\n \u003cp\u003eEngineering standards\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c4\"\u003e\n \u003cp\u003eREST\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colname=\"c1\"\u003e\n \u003cp\u003eTechnical\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c2\"\u003e\n \u003cp\u003eStackExchange\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c3\"\u003e\n \u003cp\u003eDeveloper Q\u0026amp;A\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c4\"\u003e\n \u003cp\u003eREST\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colname=\"c1\"\u003e\n \u003cp\u003eTechnical\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c2\"\u003e\n \u003cp\u003eGitHub\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c3\"\u003e\n \u003cp\u003eOpen source code\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c4\"\u003e\n \u003cp\u003eREST\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colname=\"c1\"\u003e\n \u003cp\u003eTechnical\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c2\"\u003e\n \u003cp\u003eSpringer\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c3\"\u003e\n \u003cp\u003eScientific journals\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c4\"\u003e\n \u003cp\u003eREST\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colname=\"c1\"\u003e\n \u003cp\u003eBooks\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c2\"\u003e\n \u003cp\u003eGoogle Books\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c3\"\u003e\n \u003cp\u003eDigitized books\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c4\"\u003e\n \u003cp\u003eREST\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colname=\"c1\"\u003e\n \u003cp\u003eBooks\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c2\"\u003e\n \u003cp\u003eOpen Library\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c3\"\u003e\n \u003cp\u003eUniversal catalog\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c4\"\u003e\n \u003cp\u003eREST\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colname=\"c1\"\u003e\n \u003cp\u003eArchives\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c2\"\u003e\n \u003cp\u003eInternet Archive\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c3\"\u003e\n \u003cp\u003eWeb history\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c4\"\u003e\n \u003cp\u003eREST\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colname=\"c1\"\u003e\n \u003cp\u003eGeneral\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c2\"\u003e\n \u003cp\u003eDuckDuckGo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c3\"\u003e\n \u003cp\u003eWeb search fallback\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c4\"\u003e\n \u003cp\u003eREST\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colname=\"c1\"\u003e\n \u003cp\u003eGeneral\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c2\"\u003e\n \u003cp\u003eWikipedia\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c3\"\u003e\n \u003cp\u003eReference articles\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c4\"\u003e\n \u003cp\u003eREST\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colname=\"c1\"\u003e\n \u003cp\u003eAI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c2\"\u003e\n \u003cp\u003eGoogle Gemini\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c3\"\u003e\n \u003cp\u003eAuthorship detection\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c4\"\u003e\n \u003cp\u003eREST\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n \u003c/div\u003e\n \u003cul\u003e\n \u003cli\u003e\n \u003cp\u003eOpen Scholarly Citation Indexes: OpenAlex, CrossRef, Semantic Scholar\u003c/p\u003e\n \u003c/li\u003e\n \u003cli\u003e\n \u003cp\u003eRepository Aggregators: CORE, arXiv, Europe PMC\u003c/p\u003e\n \u003c/li\u003e\n \u003cli\u003e\n \u003cp\u003eTechnical \u0026amp; Code Repositories: GitHub, StackExchange, IEEE Xplore\u003c/p\u003e\n \u003c/li\u003e\n \u003cli\u003e\n \u003cp\u003eGeneral Web \u0026amp; Book Indices: Google Books, Wikipedia, Internet Archive\u003c/p\u003e\n \u003c/li\u003e\n \u003c/ul\u003e\n \u003cp\u003eTo manage this high-volume concurrent querying without hitting browser Cross-Origin Resource Sharing (CORS) limits, we implemented a lightweight serverless proxy using Vercel Edge Functions. This proxy handles authentication, request rate-limiting, and response normalization, presenting a unified JSON schema to the frontend client.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec9\" class=\"Section2\"\u003e\n \u003ch2\u003e2.3 Algorithmic Core\u003c/h2\u003e\n \u003cp\u003eWe implemented three distinct algorithmic approaches for multi-modal plagiarism detection (Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e).\u003c/p\u003e\n \u003cdiv class=\"gridtable\"\u003e\u0026nbsp;\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e\n \u003ccaption language=\"En\"\u003e\n \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e\n \u003cdiv class=\"CaptionContent\"\u003e\n \u003cp\u003eComparison of detection algorithms implemented in PlagiarismGuard with their specific use cases and accuracy characteristics.\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003ccolgroup cols=\"5\"\u003e\u003c/colgroup\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\" colname=\"c1\"\u003e\n \u003cp\u003eAlgorithm\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\" colname=\"c2\"\u003e\n \u003cp\u003eUse Case\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\" colname=\"c3\"\u003e\n \u003cp\u003eAccuracy\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\" colname=\"c4\"\u003e\n \u003cp\u003eSpeed\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\" colname=\"c5\"\u003e\n \u003cp\u003eResilience\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colname=\"c1\"\u003e\n \u003cp\u003eN-Gram Shingling\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c2\"\u003e\n \u003cp\u003eText similarity\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c3\"\u003e\n \u003cp\u003eHigh\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c4\"\u003e\n \u003cp\u003eFast\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c5\"\u003e\n \u003cp\u003eWord reordering, synonyms\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colname=\"c1\"\u003e\n \u003cp\u003eTF-IDF Cosine\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c2\"\u003e\n \u003cp\u003eTopical matching\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c3\"\u003e\n \u003cp\u003eMedium\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c4\"\u003e\n \u003cp\u003eFast\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c5\"\u003e\n \u003cp\u003eParaphrasing\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colname=\"c1\"\u003e\n \u003cp\u003eWinnowing\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c2\"\u003e\n \u003cp\u003eCode plagiarism\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c3\"\u003e\n \u003cp\u003eHigh\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c4\"\u003e\n \u003cp\u003eMedium\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c5\"\u003e\n \u003cp\u003eVariable renaming, whitespace\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colname=\"c1\"\u003e\n \u003cp\u003ePerceptual Hash\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c2\"\u003e\n \u003cp\u003eImage detection\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c3\"\u003e\n \u003cp\u003eHigh\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c4\"\u003e\n \u003cp\u003eFast\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c5\"\u003e\n \u003cp\u003eResize, compression, format\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n \u003c/div\u003e\n \u003cp\u003e\u003cstrong\u003eNote:\u0026nbsp;\u003c/strong\u003eResilience indicates types of modifications the algorithm can detect through.\u003c/p\u003e\n \u003cdiv id=\"Sec10\" class=\"Section3\"\u003e\n \u003ch2\u003e2.3.1 Text Similarity: N-Gram Shingling\u003c/h2\u003e\n \u003cp\u003eFor rapid and robust text similarity detection, we implemented the N-Gram Shingling algorithm (Broder et al., \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e1997\u003c/span\u003e). The document D is converted into a set of contiguous subsequences (shingles) of length n (where n\u0026thinsp;=\u0026thinsp;5 to 9 words). The similarity between a suspect document A and a source document B is calculated using the Jaccard Coefficient:\u003c/p\u003e\n \u003cp\u003eJ(A,B) = |S(A) \u0026cap; S(B)| / |S(A) \u0026cup; S(B)|\u003c/p\u003e\n \u003cp\u003eThis approach is computationally efficient and resistant to minor rearrangements of words, making it superior to simple substring matching.\u003c/p\u003e\n \u003c/div\u003e\n \u003cdiv id=\"Sec11\" class=\"Section3\"\u003e\n \u003ch2\u003e2.3.2 Code Plagiarism: Winnowing\u003c/h2\u003e\n \u003cp\u003eRecognizing the rise of Computer Science education, PlagiarismGuard includes a dedicated code analysis module. We utilized the Winnowing algorithm (Schleimer et al., \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e2003\u003c/span\u003e), a local fingerprinting algorithm utilized by MOSS (Measure of Software Similarity). The algorithm tokenizes source code (ignoring comments and whitespace), generates a stream of hashes, and selects a subset of these hashes (fingerprints) based on a sliding window. This guarantees that any match of text at least as long as the window size is detected, offering robust performance against variable renaming and reformatting attacks.\u003c/p\u003e\n \u003c/div\u003e\n \u003cdiv id=\"Sec12\" class=\"Section3\"\u003e\n \u003ch2\u003e2.3.3 Visual Similarity: Perceptual Hashing\u003c/h2\u003e\n \u003cp\u003eTo detect image plagiarism\u0026mdash;critical for figures in scientific manuscripts\u0026mdash;we implemented a perceptual hashing (pHash) algorithm (Zauner, \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2010\u003c/span\u003e). Unlike cryptographic hashes (like MD5) where a single pixel change results in a completely different hash, perceptual hashes adapt so that similar images produce similar hash values. We use the Discrete Cosine Transform (DCT) to generate a \u0026quot;fingerprint\u0026quot; of the image\u0026apos;s low-frequency components, allowing the system to detect images that have been resized, compressed, or slightly color-corrected.\u003c/p\u003e\n \u003c/div\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec13\" class=\"Section2\"\u003e\n \u003ch2\u003e2.4 AI Authorship Analysis\u003c/h2\u003e\n \u003cp\u003ePlagiarismGuard adheres to a \u0026quot;neutral platform\u0026quot; philosophy regarding AI detection. Instead of building a proprietary black-box classifier, it integrates with state-of-the-art external Large Language Models (LLMs) via API. Users can bring their own API keys (BYOK) for Google Gemini, OpenAI, or Anthropic Claude. The system constructs specialized prompts designed to analyze linguistic features\u0026mdash;such as perplexity, burstiness, and semantic coherence\u0026mdash;to estimate the likelihood of AI generation. This approach ensures the tool remains up-to-date with the latest foundational models, unlike static classifiers that quickly become obsolete.\u003c/p\u003e\n\u003c/div\u003e"},{"header":"3. Results","content":"\u003cdiv id=\"Sec15\" class=\"Section2\"\u003e\n \u003ch2\u003e3.1 Implementation and Deployment\u003c/h2\u003e\n \u003cp\u003eThe complete system has been open-sourced on GitHub and deployed as a live service reachable via two mirrors to ensure redundancy: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://plagiarism-checker-web-app.vercel.app/\u003c/span\u003e\u003c/span\u003e (Primary) and \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://plagiarismguardpro.netlify.app/\u003c/span\u003e\u003c/span\u003e (Secondary). The application features a responsive Material Design interface that adapts to desktop, tablet, and mobile workflows. As a PWA, it supports offline installation, allowing users in low-bandwidth environments to load the interface and queue documents for analysis.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec16\" class=\"Section2\"\u003e\n \u003ch2\u003e3.2 Functional Performance \u0026amp; Reporting\u003c/h2\u003e\n \u003cp\u003eThe platform successfully processes standard manuscript formats (.txt, .docx, .pdf). The main analysis interface provides real-time feedback and metrics (Fig. \u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e2\u003c/span\u003e). Upon analysis, users receive a comprehensive \u0026quot;plagiarism report\u0026quot; certificate (Fig. \u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e3\u003c/span\u003e). This report includes:\u003c/p\u003e\n \u003cul\u003e\n \u003cli\u003e\n \u003cp\u003eSimilarity Score Validation: A global percentage representing the total overlap with external sources.\u003c/p\u003e\n \u003c/li\u003e\n \u003cli\u003e\n \u003cp\u003eSource Breakdown: An itemized list of all detected sources, categorized by type (Academic Journal, Website, Repository).\u003c/p\u003e\n \u003c/li\u003e\n \u003cli\u003e\n \u003cp\u003eTextual Highlighting: Direct side-by-side comparison of the user\u0026apos;s text against the matched external source, utilizing color-coded highlights for easy visual auditing.\u003c/p\u003e\n \u003c/li\u003e\n \u003cli\u003e\n \u003cp\u003eVerification Metadata: A QR code and cryptographic hash of the report, ensuring the certificate\u0026apos;s authenticity and preventing tampering.\u003c/p\u003e\n \u003c/li\u003e\n \u003c/ul\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec17\" class=\"Section2\"\u003e\n \u003ch2\u003e3.3 Performance Benchmarking\u003c/h2\u003e\n \u003cp\u003eTo evaluate system efficiency, we benchmarked PlagiarismGuard against a standard 3,000-word academic manuscript. On a standard broadband connection:\u003c/p\u003e\n \u003cul\u003e\n \u003cli\u003e\n \u003cp\u003eInitialization Time: \u0026lt; 1.5 seconds to load the core application shell.\u003c/p\u003e\n \u003c/li\u003e\n \u003cli\u003e\n \u003cp\u003eAPI Aggregation Latency: The parallel execution of 16 API queries typically resolves within 4\u0026ndash;8 seconds.\u003c/p\u003e\n \u003c/li\u003e\n \u003cli\u003e\n \u003cp\u003eProcessing Throughput: The client-side tokenization and shingling engine processes approximately 10,000 words per second on an average consumer laptop (Intel i5 equivalent).\u003c/p\u003e\n \u003c/li\u003e\n \u003c/ul\u003e\n \u003cp\u003eThis performance profile confirms that client-side JavaScript execution is more than sufficient for real-time document analysis, challenging the assumption that such heavy lifting requires dedicated server clusters.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec18\" class=\"Section2\"\u003e\n \u003ch2\u003e3.4 Comparison with Commercial Tools\u003c/h2\u003e\n \u003cp\u003eTable\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e presents a comparative analysis. While commercial tools maintain larger proprietary databases, PlagiarismGuard offers comparable algorithmic sophistication with significant cost advantages. Note: Database coverage is simulated via aggregate API access.\u003c/p\u003e\n \u003cdiv class=\"gridtable\"\u003e\u0026nbsp;\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e\n \u003ccaption language=\"En\"\u003e\n \u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e\n \u003cdiv class=\"CaptionContent\"\u003e\n \u003cp\u003eComparative analysis of PlagiarismGuard against leading commercial plagiarism detection tools.\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003ccolgroup cols=\"5\"\u003e\u003c/colgroup\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\" colname=\"c1\"\u003e\n \u003cp\u003eFeature\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\" colname=\"c2\"\u003e\n \u003cp\u003ePlagiarismGuard\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\" colname=\"c3\"\u003e\n \u003cp\u003eTurnitin\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\" colname=\"c4\"\u003e\n \u003cp\u003eiThenticate\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\" colname=\"c5\"\u003e\n \u003cp\u003eCopyscape\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colname=\"c1\"\u003e\n \u003cp\u003eCost\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c2\"\u003e\n \u003cp\u003eFree (MIT License)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c3\"\u003e\n \u003cp\u003e\u003cspan\u003e$\u003c/span\u003e3-15k/year\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c4\"\u003e\n \u003cp\u003e\u003cspan\u003e$\u003c/span\u003e50k+/year\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c5\"\u003e\n \u003cp\u003e\u003cspan\u003e$\u003c/span\u003e0.05/page\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colname=\"c1\"\u003e\n \u003cp\u003eDatabase Size\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c2\"\u003e\n \u003cp\u003e16 Open APIs\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c3\"\u003e\n \u003cp\u003e100B+ pages\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c4\"\u003e\n \u003cp\u003e200M+ articles\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c5\"\u003e\n \u003cp\u003eWeb Index\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colname=\"c1\"\u003e\n \u003cp\u003eAI Detection\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c2\"\u003e\n \u003cp\u003eBYOK (Gemini/GPT)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c3\"\u003e\n \u003cp\u003eProprietary\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c4\"\u003e\n \u003cp\u003eProprietary\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c5\"\u003e\n \u003cp\u003eNone\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colname=\"c1\"\u003e\n \u003cp\u003ePrivacy\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c2\"\u003e\n \u003cp\u003eClient-side only\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c3\"\u003e\n \u003cp\u003eCloud storage\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c4\"\u003e\n \u003cp\u003eCloud storage\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c5\"\u003e\n \u003cp\u003eCloud storage\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n \u003c/div\u003e\n \u003cp\u003e\u003cstrong\u003eNote:\u0026nbsp;\u003c/strong\u003eCommercial pricing is approximate and varies by institution size and contract terms.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec19\" class=\"Section2\"\u003e\n \u003ch2\u003e3.5 Mobile Accessibility \u0026amp; PWA\u003c/h2\u003e\n \u003cp\u003eBeing a Progressive Web Application, PlagiarismGuard can be installed on mobile devices for quick checks (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e4\u003c/span\u003e). The responsive interface adapts seamlessly from desktop to smartphone, with touch-optimized 44px minimum tap targets. This zero-cost accessibility directly addresses the \u0026quot;integrity divide\u0026quot; faced by LMIC institutions.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec20\" class=\"Section2\"\u003e\n \u003ch2\u003e3.6 User Privacy\u003c/h2\u003e\n \u003cp\u003eA key result of this architectural approach is \u0026quot;Privacy by Design.\u0026quot; We verified via network packet inspection that the full manuscript text body is never transmitted to our servers. Only search queries (3\u0026ndash;5 word phrases) and statistical metadata are sent to the API proxy. This offers a level of intellectual property protection that is structurally impossible for centralized cloud-based competitors to match.\u003c/p\u003e\n\u003c/div\u003e"},{"header":"4. Discussion","content":"\u003cdiv id=\"Sec22\" class=\"Section2\"\u003e \u003ch2\u003e4.1 Empowering the \"Long Tail\" of Academia\u003c/h2\u003e \u003cp\u003eThe primary contribution of PlagiarismGuard is the democratization of academic verification. By removing the cost barrier, we empower the \"long tail\" of the academic community\u0026mdash;independent scholars, students in underfunded institutions, and researchers in developing nations. This aligns with the United Nations Sustainable Development Goal 4 (Quality Education) by ensuring equitable access to technical tools required for academic success.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec23\" class=\"Section2\"\u003e \u003ch2\u003e4.2 The Role of Open Source in EdTech\u003c/h2\u003e \u003cp\u003ePlagiarismGuard demonstrates the viability of Open Source Software (OSS) in the educational technology sector. Unlike closed proprietary systems, OSS allows institutions to audit the matching algorithms for bias (e.g., checking if the tool adequately indexes non-English repositories). Furthermore, the extensibility of our plugin architecture allows developers to add new data sources (such as regional language databases) without waiting for a commercial vendor to prioritize them.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec24\" class=\"Section2\"\u003e \u003ch2\u003e4.3 Limitations\u003c/h2\u003e \u003cp\u003eWhile promising, our approach has limitations. First, reliance on public APIs means coverage is strictly limited to Open Access content; paywalled journals indexed deeply by Turnitin may occasionally be missed, though abstract-level matching often mitigates this. Second, the \"BYOK\" model for AI detection shifts the cost of API tokens to the user, although this cost is negligible compared to annual subscriptions. Finally, as a client-side tool, extremely large documents (e.g., whole theses\u0026thinsp;\u0026gt;\u0026thinsp;100MB) may face browser memory constraints.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec25\" class=\"Section2\"\u003e \u003ch2\u003e4.4 Future Directions\u003c/h2\u003e \u003cp\u003eFuture development will focus on three areas: (1) Developing a Learning Management System (LMS) plugin (LTI standard) for seamless integration with Moodle and Canvas; (2) Enhancing cross-language plagiarism detection using vector embeddings; and (3) Creating a federated database network where institutions can opt-in to share anonymized fingerprints, creating a community-sourced comparison database.\u003c/p\u003e \u003c/div\u003e"},{"header":"5. Conclusions","content":"\u003cp\u003ePlagiarismGuard represents a paradigm shift from \"plagiarism policing\" to \"integrity empowerment.\" By providing a free, transparent, and privacy-respecting tool, we offer a viable alternative to the commercial oligopoly in academic verification. We invite the educational community to adopt, fork, and improve this platform to ensure that financial resources never determine an institution's ability to uphold academic standards.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003eFunding: No funding was received for this work.\u003c/p\u003e\n\u003cp\u003eConflicts of Interest: The author declares no conflicts of interest. The software is free and open-source.\u003c/p\u003e\n\u003cp\u003eAvailability of Data and Material: Source code is available at GitHub (https://github.com/hssling/Plagiarism_Checker_Web_App) under the MIT License. Live deployment is accessible at https://plagiarism-checker-web-app.vercel.app/ and https://plagiarismguardpro.netlify.app/.\u003c/p\u003e\n\u003cp\u003eEthics Approval: Not applicable. This study describes software development and does not involve human subjects or personal data.\u003c/p\u003e\n\u003cp\u003eClinical trial registration: Not applicable. This study describes software development and does not involve human subjects or personal data Hence clinical trial registration number and date of registration are not applicable.\u003c/p\u003e\n\u003cp\u003eConsent to Publish declaration: not applicable\u003c/p\u003e\n\u003cp\u003eConsent to Participate declaration: not applicable\u0026nbsp;\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eBretag, T., \u0026amp; Mahmud, S. (2009). Self-plagiarism or appropriate textual re-use? Journal of Academic Ethics, 7(3), 193-205. https://doi.org/10.1007/s10805-009-9092-1\u003c/li\u003e\n\u003cli\u003eBroder, A. Z., Glassman, S. C., Manasse, M. S., \u0026amp; Zweig, G. (1997). Syntactic clustering of the web. Computer Networks and ISDN Systems, 29(8-13), 1157-1166. https://doi.org/10.1016/S0169-7552(97)00031-7\u003c/li\u003e\n\u003cli\u003eCoughlan, S. (2019). Universities spend millions on plagiarism detection. BBC News. Retrieved from https://www.bbc.com/news/education-50420455\u003c/li\u003e\n\u003cli\u003eFolt\u0026yacute;nek, T., Meuschke, N., \u0026amp; Gipp, B. (2020). Academic plagiarism detection: A systematic literature review. ACM Computing Surveys, 52(6), 1-42. https://doi.org/10.1145/3345317\u003c/li\u003e\n\u003cli\u003eGoogle. (2025). Gemini API Documentation. Retrieved from https://ai.google.dev/docs\u003c/li\u003e\n\u003cli\u003eICMJE. (2024). Recommendations for the Conduct, Reporting, Editing, and Publication of Scholarly Work in Medical Journals. Retrieved from https://www.icmje.org/recommendations/\u003c/li\u003e\n\u003cli\u003eMozilla Laboratory. (2024). Cross-Origin Resource Sharing (CORS). MDN Web Docs. Retrieved from https://developer.mozilla.org/en-US/docs/Web/HTTP/CORS\u003c/li\u003e\n\u003cli\u003ePriem, J., Piwowar, H., \u0026amp; Orr, R. (2022). OpenAlex: A fully-open index of scholarly works, authors, venues, institutions, and concepts. arXiv preprint arXiv:2205.01833\u003c/li\u003e\n\u003cli\u003ePupovac, V., \u0026amp; Fanelli, D. (2015). Scientists admitting to plagiarism: A meta-analysis of surveys. Science and Engineering Ethics, 21(5), 1331-1352\u003c/li\u003e\n\u003cli\u003eResnik, D. B., \u0026amp; Master, Z. (2013). Policies and initiatives aimed at addressing research misconduct in high-income countries. PLoS Medicine, 10(3), e1001406\u003c/li\u003e\n\u003cli\u003eSalton, G., \u0026amp; Buckley, C. (1988). Term-weighting approaches in automatic text retrieval. Information Processing \u0026amp; Management, 24(5), 513-523\u003c/li\u003e\n\u003cli\u003eSchleimer, S., Wilkerson, D. S., \u0026amp; Aiken, A. (2003). Winnowing: Local algorithms for document fingerprinting. Proceedings of the 2003 ACM SIGMOD International Conference on Management of Data, 76-85\u003c/li\u003e\n\u003cli\u003eTennant, J. P., Waldner, F., Jacques, D. C., et al. (2016). The academic, economic and societal impacts of Open Access: an evidence-based review. F1000Research, 5, 632\u003c/li\u003e\n\u003cli\u003eTurnitin. (2025). iThenticate: Professional Plagiarism Prevention. Retrieved from https://www.ithenticate.com/\u003c/li\u003e\n\u003cli\u003eWAME. (2023). Recommendations on Publication Ethics Policies for Medical Journals. World Association of Medical Editors. Retrieved from https://wame.org/\u003c/li\u003e\n\u003cli\u003eW3C. (2023). Web Application Manifest. W3C Working Draft. Retrieved from https://www.w3.org/TR/appmanifest/\u003c/li\u003e\n\u003cli\u003eWeber-Wulff, D. (2014). False Feathers: A Perspective on Academic Plagiarism. Berlin: Springer\u003c/li\u003e\n\u003cli\u003eWeber-Wulff, D., Anohina-Naumeca, A., Bjelobaba, S., et al. (2023). Testing of detection tools for AI-generated text. International Journal for Educational Integrity, 19(1), 26\u003c/li\u003e\n\u003cli\u003eZauner, C. (2010). Implementation and benchmarking of perceptual image hash functions [Master\u0026apos;s thesis]. Hagenberg, Austria: Upper Austria University of Applied Sciences.\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"discover-artificial-intelligence","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"diai","sideBox":"Learn more about [Discover Artificial Intelligence](https://www.springer.com/44163)","snPcode":"","submissionUrl":"","title":"Discover Artificial Intelligence","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Discover Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"plagiarism detection, academic integrity, educational technology, open-source software, generative AI, digital equity, higher education","lastPublishedDoi":"10.21203/rs.3.rs-8988484/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8988484/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eAcademic integrity is facing an unprecedented crisis in the digital era, exacerbated by the proliferation of \"paper mills\" and the rapid adoption of sophisticated Generative AI tools. While commercial detection solutions like iThenticate and Turnitin have established themselves as industry standards, their prohibitive cost models create a significant \"integrity divide,\" disproportionately affecting institutions in low- and middle-income countries (LMICs) and independent researchers. This unavailability of affordable verification tools risks compromising the quality of scientific output from resource-constrained regions. Addressing this disparity, we present PlagiarismGuard, a free, open-source, client-side academic integrity verification platform. Built on a modern serverless architecture, PlagiarismGuard aggregates search results from 16 open academic databases—including OpenAlex, Semantic Scholar, and CrossRef—to perform comprehensive similarity analysis without requiring institutional subscriptions. The system implements advanced text fingerprinting algorithms (n-gram shingling), code plagiarism detection (Winnowing), and perceptual image hashing, while also integrating a \"Bring Your Own Key\" (BYOK) model for AI-powered authorship analysis using models like Google Gemini and OpenAI GPT-4. We detail the system's architecture, privacy-preserving client-side processing model, and performance benchmarks, demonstrating that open-source alternatives can provide robust plagiarism detection capabilities comparable to commercial tools. By releasing PlagiarismGuard under the MIT License, we aim to democratize access to academic integrity tools and foster a community-driven approach to combating research misconduct.\u003c/p\u003e","manuscriptTitle":"PlagiarismGuard: Democratizing Academic Integrity with a Free, Open-Source, Multi-API Plagiarism Detection Ecosystem","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-04-10 16:57:30","doi":"10.21203/rs.3.rs-8988484/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"editorInvitedReview","content":"","date":"2026-05-09T04:58:29+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"88552711150367194350430628059321457551","date":"2026-05-07T03:10:17+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-04-19T19:27:25+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"336198127978171728729015980848478473960","date":"2026-04-14T11:13:10+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-04-09T11:48:30+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"65493629969903519128257227978263729271","date":"2026-04-09T11:04:52+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"192606656342047096578952162511194585244","date":"2026-04-08T13:35:11+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"156278168832800333738239232205983009284","date":"2026-04-06T15:23:24+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2026-04-06T10:38:59+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2026-03-03T07:21:22+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2026-03-03T07:20:23+00:00","index":"","fulltext":""},{"type":"submitted","content":"Discover Artificial Intelligence","date":"2026-02-27T13:09:40+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"discover-artificial-intelligence","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"diai","sideBox":"Learn more about [Discover Artificial Intelligence](https://www.springer.com/44163)","snPcode":"","submissionUrl":"","title":"Discover Artificial Intelligence","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Discover Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"802c1010-0f3d-4715-b459-b7762ae854d8","owner":[],"postedDate":"April 10th, 2026","published":true,"recentEditorialEvents":[{"type":"editorInvitedReview","content":"","date":"2026-05-09T04:58:29+00:00","index":69,"fulltext":""},{"type":"reviewerAgreed","content":"88552711150367194350430628059321457551","date":"2026-05-07T03:10:17+00:00","index":68,"fulltext":""}],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2026-04-10T16:57:30+00:00","versionOfRecord":[],"versionCreatedAt":"2026-04-10 16:57:30","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8988484","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8988484","identity":"rs-8988484","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}