W14-01: Uncovering Definition Coverage in the OBO Foundry Ontologies

TitleW14-01: Uncovering Definition Coverage in the OBO Foundry Ontologies
Publication TypeConference Paper
Year of Publication2016
AuthorsSchlegel DR, Seppälä S, Elkin PL
Conference NameInternational Conference on Biomedical Ontology and BioCreative (ICBO BioCreative 2016)
Date Published11/30/16
PublisherCEUR-ws.org Volume 1747
Conference LocationCorvallis, Oregon, USA
Other NumbersVol-1747|urn:nbn:de:0074-1747-1
Keywordsdefinition coverage, logical definitions, OBO Foundry, ontology, textual definitions

Definitions, both logical and textual, are an essential part of ontologies. Textual definitions help human users disambiguate and regularize their understanding and use of ontology terms to achieve intra- and inter-personal consistency and avoid errors, for example, when annotating scientific data, integrating databases with an ontology, or importing terms into other ontologies. Logical definitions are needed, among other things, for checking the consistency of the ontology and carrying out inferences, for example, over data that has been annotated with ontology terms. Despite the best efforts of ontology developers, it is not uncommon to see missing definitions. While the OBO Foundry explicitly states that its member ontologies should have a substantial fraction of their terms defined, these ontologies still often lack one or both kinds of definitions. Statistics on definition coverage in the OBO Foundry ontologies are scarce and it is difficult to tell what effectively constitutes a substantial fraction of terms in an ontology. In the present work, we examine the coverage of textual and logical definitions throughout the OBO Foundry ontologies in order to uncover the big picture and to give more detailed insight into logical definitions in these ontologies. We have found that textual definition coverage is reasonably good over the OBO Foundry ontologies (66%), but that the core ontologies exhibit a higher definition coverage (86%) than the non-core ones (64%). Logical definitions follow a similar trend, but with lower values — overall, the OBO Foundry has a 30% coverage, while core ontologies are better covered (53%) than non-core ones (28%).