Simple Models, Complex Vocabularies: Developing Controlled Vocabularies for an Interdisciplinary Collection Management System in RECODE

Situated at the intersection of distinct stakeholder communities and their objectives, collection management systems (CMS) need to integrate and mediate a wide range of demands to provide functionality, user experience, and data fit for purpose. While metadata standards, (e.g., Biodiversity Information Standards (TDWG) Darwin Core (Darwin Core Task Group 2009) and its Latimer Core (Grant et al. 2024), and Pinian Core (Plinian Core Task Group 2021) extensions) and ontologies, (e.g., the World Wide Web Consortium (W3C) Provenance Ontology (Lebo et al. 2013) or the W3C Open Digital Rights Language (Iannella et al. 2018)) provide guidance for structuring data resources and workflows, controlled vocabularies standardize and harmonize the data content in those structures. Controlled vocabularies contribute to differentiating dimensions of information present in metadata concepts and allow comprehensive, information-rich descriptions of reality by aiming to provide well-defined terms that can be clearly understood. Harmonized across scientific and applied disciplines as well as distributed data infrastructures, they contribute to data interoperability, findability, and reusability, and thus to the basis for data sharing and the automation of work processes. Instead of introducing challenges for users, the presentation of context-specific subsets of terms for manual selection as well as automation of context-deducible entries can improve user experiences, work environment efficiency, and (meta)data comprehensiveness. This shifts infrastructure development to an additional layer of rules and constraints (policies) that determine interface dynamics and data validation. Setting these theoretical considerations to the test of practice, we are sharing our experiences and insights gained during the development and implementation of the new collection management system by the RECODE (Rethinking Collections Data Ecosystems) program at the Natural History Museum, London. Controlled vocabularies and their terms constitute a major component in the CMS data model. They present challenges due to their context-specificity and hierarchical nature, for which solutions need to be found. Daily work with controlled vocabularies requires extensive documentation with functionality for creating and tracking provenance, relationships, and mappings, as well as for versioning. There is a need for open, shared repositories and work environments that foster the versatile, user-driven develo