MDG Conference 2026 - Authentic and Artificial Cataloguing

🎤 Presentation given by Hannes Lowagie at the Metadata & Discovery Group Conference 2026. 📑 The slides are available at https://doi.org/10.5281/zenodo.19554062. 📒 An article version is also available in issue 216 of Catalogue & Index: https://journals.cilip.org.uk/catalog... 📝 Abstract Cataloguers and metadata professionals stand at the threshold of a new era, one where local cataloguing practices must not only be transparent and consistent but also machine actionable. This presentation introduces a metadata strategy that reimagines how cataloguing rules are authored, shared, and used—both by people and machines—laying foundational work for broader metadata interoperability and even the long-envisioned goal of Universal Bibliographic Control (UBC). Traditionally, cataloguing guidelines have been published in formats suited only for human consumption: print manuals, PDF documents, or static HTML pages. KBR, the royal library of Belgium, has moved beyond these formats to develop a modular, machine-readable metadata ruleset that takes the form of a structured JSON file. This JSON file is now the single source of truth for our cataloguing rules, providing data to two complementary outputs: 1. A human-readable HTML interface that dynamically renders rule descriptions and logic using the JSON backend, and 2. A machine-readable interface that enables integration with generative AI tools and other metadata processing systems by sending the same JSON content as part of prompt engineering workflows. This “triangle model” — JSON as the backbone, HTML as the human interface, and prompt ready data as the machine interface — represents a novel, scalable solution to metadata consistency and authority control in both traditional and AI-assisted cataloguing environments. The implications are significant. If libraries and institutions adopt similar machine-readable frameworks for their local rules, the path to aligning and ultimately unifying metadata practices becomes more achievable. Our approach goes beyond standard creation or adoption — it promotes harmonization of logic, not just terminology. By enabling AI tools to respect local policies while operating at scale, we ensure that future metadata work maintains fidelity, transparency, and context. This presentation will offer a walkthrough of the JSON schema and discuss technical considerations, and demonstrate how this system is already being used in conjunction with generative AI tools. It concludes with a call for collective action: imagine a future in which every institution’s rules are machine-readable and designed to interoperate. Why not aim not only for compatibility, but for shared cataloguing logic—a collaborative metadata ecosystem that supports both human judgment and machine efficiency? This talk is especially relevant to metadata specialists, cataloguers, system librarians, and LIS students interested in the intersections of metadata governance, automation, and the evolving role of cataloguing professionals in AI-enhanced environments. 🗣️ Hannes Lowagie 🏛️ Koninklijke Bibliotheek van België - Bibliothèque royale de Belgique Hannes Lowagie (he/him) is head of the Agency for Bibliographic Information in Royal Library of Belgium (KBR), the National Library of Belgium. He is currently working on the transition to RDA, Linked Open Data and the use of AI-technologies in the cataloguing process. He has a PhD in (Medieval) History (University of Ghent, 2012).