The structure of knowledge organization systems (KOSs) – domain vocabularies, thesauri, terminologies, classification systems, and ontologies – follows different architectural principles and semantic theories. However, many use cases require their integrated use in a given domain. Building a common framework for KOSs is then a prerequisite for any principled account of their use when data annotated by different KOSs should be integrated. We propose an approach rooted in formal ontology, the aim of which is to harmonize the description of the domain itself with the description of the representational artifacts that claim to organize and represent knowledge of this domain. We propose a transparent framework for describing KOSs with a focus on the biomedical domain. Using comprehensive and consistent terminology, we formalize what KOSs represent by introducing KOSonto, an ontology that characterizes representational artifacts on the one hand and describes the relationships to their referents in the domain of application on the other hand. KOSonto uses OWL-DL axioms and is built under BFO and IAO. It accounts for a range of elements that are characteristic of different kinds of KOSs. We illustrate how KOSonto can be used to describe typical biomedical KOSs such as ICD-10, SNOMED CT, and MeSH. Further work will improve the alignment of KOSonto to foundational ontologies and apply this framework to optimize the creation, use, and reuse of mappings between heterogeneous KOSs.< Réduire