Digital Archiving of Sacred Texts: AI-Driven Preservation and Accessibility

Thesis Abstract

Digital Archiving of Sacred Texts seeks to address the critical challenge of preserving vulnerable religious manuscripts and oral traditions in the digital era while ensuring broad, equitable access for scholars and communities. The study situates itself within the intersection of digital humanities, information science, and religious/cultural studies, responding to risks of material degradation, geopolitical instability, and the marginalization of minority Scriptural heritages in mainstream digital repositories. The aim is to develop and evaluate an AI-driven archiving framework that enables high-fidelity digitization, automated metadata generation, multilingual transcription, and smart access controls aligned with cultural and theological sensitivities. Specific objectives include (1) designing an end-to-end digitization workflow that integrates high-resolution imaging, optical character recognition (OCR) for printed texts and neural handwriting recognition for manuscripts, (2) implementing a multimedia metadata schema and ontology grounded in FRBR-LRM concepts augmented by religious studies vocabularies to support precise retrieval, (3) deploying AI-based automated quality assurance and error correction pipelines to improve transcription accuracy across scripts (e.g., Devanagari, Arabic, Ethiopic, Greek), (4) evaluating accessibility and discoverability through user-centered metrics across diverse user groups including clergy, scholars, and community archivists, and (5) assessing ethical, legal, and community governance considerations for controlled access to sensitive texts. The methodology adopts a mixed-methods design anchored in the praxeology of digital preservation and the technology acceptance literature. The population comprises 24 advanced archival repositories and 12 community libraries holding sacred texts from five religious traditions. A purposive sample of 40 digitization projects and 120 practitioners (librarians, archivists, theologians, and IT specialists) will be studied. Data collection instruments include (a) structured questionnaires to gauge perceived usefulness, ease of use, and trust in AI-assisted workflows; (b) semi-structured interviews to explore governance, consent, and cultural sensitivities; (c) document analysis of existing digitization standards and licensing regimes; (d) pilot datasets consisting of 600 scanned pages and 120 audio recordings across five languages for OCR/ASR evaluation; and (e) usability test tasks with 60 participants to assess search effectiveness and interface accessibility. Validity and reliability will be established through triangulation, pilot testing of instruments (n=20) and inter-coder agreement for qualitative data (Cohen’s kappa > 0.75). Data analysis combines quantitative and qualitative techniques regression analysis and structural equation modeling (SEM) to test the relationships between system quality, information quality, and user adoption; ANOVA to compare performance across scripts; precision, recall, and F-measure metrics for OCR/ASR outputs; and thematic analysis within an applied framework drawing on Hassan and Ali’s theory of religious information behavior. A conceptual model will be proposed and tested to capture the interaction between AI-driven digitization processes, metadata richness, and user access profiles, integrating theories of Diffusion of Innovations and Information Behavior in religious contexts. Anticipated findings indicate that AI-assisted workflows can achieve transcription accuracy improvements of up to 92% for print texts and 78% for manuscripts after domain-specific fine-tuning, with metadata completeness increasing by 35% and search precision by 28% compared to baseline manual processes. The study expects to reveal critical trade-offs between openness and cultural control, demonstrating that tiered access and provenance-tracking significantly enhance trust and uptake among community stakeholders. Contributions to knowledge include a replicable AI-enabled archiving framework tailored for sacred texts, an interoperable metadata ontology bridging religious studies and library science, and an evidence-based governance model for ethical access. The research will inform best practices for future digitization initiatives and offer scalable guidelines for preserving intangible aspects such as chant, recitation, and ritual context embedded in audio and visual materials. The conclusion emphasizes the viability of AI-assisted digital archiving as a means to safeguard sacred textual heritage while expanding scholarly and community engagement, with practical recommendations for repository designers, funders, and policy-makers to balance preservation, accessibility, and cultural sovereignty.