The Quad Economy of a Semantic Web Ontology Repository

Quad stores have a number of features that make them very attractive for Semantic Web applications. Quad stores use Named Graphs to give contextual information to RDF graphs. Developers can use these contexts to incorporate meta-data such as provenance and versioning. The OWL language specification provides the owl:imports construct as one of the key ways to reuse the ontologies on the Semantic Web. When one ontology imports another through the owl:imports statement, all axioms from the imported ontology are brought to the source ontology. When implementing an online ontology repository, we have explored a number of different approaches to reflect these imports and contextual information in a quad store. These approaches have a direct impact on storage requirements, query articulation, and reusability. This paper describes different models to represent contextual information, ontology imports and versioning. We have performed the experiments in the context of storing ontologies from the BioPortal ontology repository. This repository has 304 ontologies, with multiple versions for many of them. We extract metrics on the levels of imports, and the storage and performance requirements and discuss the trade-offs among query articulation, reusability and scalability.