Authentic and artificial cataloguing

machine-readable cataloguing rules as a bridge to unified metadata practice

Abstract

Cataloguers and metadata professionals are entering a period in which local cataloguing rules must be not only transparent and consistent, but also usable by machines. This article examines a metadata strategy developed at KBR, Royal Library of Belgium that transforms traditional cataloguing guidance into a modular, machine-readable ruleset structured in JSON. Rather than maintaining separate documentation for staff and systems, the JSON file functions as a single source of truth that supports two outputs: a dynamic HTML interface for human users and a machine-readable interface that can be integrated into generative AI workflows and other metadata processing environments.

The article explores how this “triangle model” - JSON as the core structure, HTML as the human interface, and prompt-ready data as the machine interface - offers a scalable approach to maintaining consistency, authority control, and transparency in cataloguing practice. It also considers the wider implications for interoperability across libraries and cultural heritage institutions. If local rules are encoded in comparable machine-readable forms, alignment between institutions may move beyond shared terminology towards shared cataloguing logic. The article concludes by reflecting on how metadata professionals can shape AI-enhanced cataloguing environments while preserving professional judgement, local context, and accountability.