Abstract
National Bibliographic Agencies (NBAs) occupy a pivotal position in the global information ecosystem, curating authoritative bibliographic and authority data. However, traditional cataloging frameworks based on MARC standards inherently limit the discoverability and semantic interoperability of library resources across the broader Web. In response, several NBAs have transitioned toward Linked Open Data (LOD) paradigms to unlock bibliographic silos and facilitate cross-domain knowledge integration. This study presents a rigorous comparative analysis of three predominant Linked Data implementation architectures currently deployed by leading NBAs: Pipeline-based Batch ETL Transformation, Native Graph Knowledge Base Architectures, and Dynamic Virtualized API Middleware. Utilizing a benchmark dataset of 500,000 multi-format bibliographic records, we evaluate these strategies across four core dimensions: semantic fidelity and ontology mapping coverage, query latency under concurrent SPARQL workloads, entity reconciliation precision against external hubs (e.g., VIAF, Wikidata, and ISNI), and operational maintenance overhead. The findings reveal that while native graph architectures deliver superior semantic expressiveness and dynamic linking capabilities, hybrid virtualized architectures offer the most pragmatically viable trajectory for legacy-encumbered institutions. We synthesize these empirical results into an actionable decision-making framework to guide national libraries in formulating scalable, interoperable, and future-proof LOD strategies.
