Ontological Representation of FAIR Principles: A Blueprint for FAIRer Data Sources

Guidelines to improve the Findability, Accessibility, Interoperability, and Reuse of datasets, known as FAIR principles, were introduced in 2016 to enable machines to perform automatic actions on a variety of digital objects, including datasets. Since then, the principles have been widely adopted by data creators and users worldwide with the ‘FAIR’ acronym becoming a common part of the vocabulary of data scientists. However, there is still some controversy on how datasets should be interpreted since not all datasets that are claimed to be FAIR, necessarily follow the principles. In this research, we propose the OntoUML FAIR Principles Schema, as an ontological representation of FAIR principles for data practitioners. The work is based on OntoUML, an ontologically well-founded language for Ontology-driven Conceptual Modeling. OntoUML is a proxy for ontological analysis that has proven effective in supporting the explanation of complex domains. Our schema aims to disentangle the intricacies of the FAIR principles’ definition, by resolving aspects that are ambiguous, under-specified, recursively-specified, or implicit. The schema can be considered as a blueprint, or a template to follow when the FAIR classification strategy of a dataset must be designed. To demonstrate the usefulness of the schema, we present a practical example based on genomic data and discuss how the results provided by the OntoUML FAIR Principles Schema contribute to existing data guidelines.