Building better controlled vocabularies: Guidance from ISO 25964

Based on examples of existing controlled vocabularies and use cases, some quality issues are evaluated against recommendations given by the ISO standard and SKOS. While many validity criteria can be applied as stated in the standard, some areas of conflict come to light when attempting to retain full conformance. The model elements of thesaurus array and compound equivalence leave the implementer with some ambiguity and for the latter, an alternative to the standard is suggested. The proposals for implementing facets provide for several options, but fall short of explaining the implications of the different approaches when applied to vocabulary management and information retrieval tasks. Existing applications of a top concept relation do not always conform with the notion as defined by the standard, giving rise to conflicts between declared and inferred hierarchy relationships. It is discussed whether explicitly asserted top concept relationships can augment a thesaurus beyond what is possible by automatic evaluation of concept hierarchies. Some of these issues are illustrated using SPARQL-based queries over existing thesauri.