An Ontology-Based Source Selection for Federated Query Processing: a Case Study

Abstract

An important step in federated query execution frameworks is source selection, determining which endpoints are relevant to evaluate a given query. Source selection process happens as a separate step before the federated query by executing SPARQL ASK queries, updating catalog/index, or collecting heuristic information as a pre-processing stage, however, in domains as the Linked Open University context, these strategies involve some issues. On the other side, the DCAT metadata vocabulary enables a publisher to describe datasets and data services in a catalog using a standard model and vocabulary that facilitates their consumption. In addition, data summarizations are a lightweight form of representing crucial dataset information. Moreover, the Hydra hypermedia vocabulary along with its Hydra API Documentation allow describing RDF Web APIs facilitating the automation of the client-server communication. This work focuses on using the former semantic vocabularies, along with a context-based unified well-accepted vocabulary, in favor of facilitating the source selection process. In order to explain our proposal, a case study in the Linked Open University context was presented. The case study showed that our proposal allows to select the right sources per triple pattern without further processing complexity as the usage of SPARQL ASK queries and, in turn, it is tailored not only to established query interfaces as SPARQL endpoints, but also to new query interfaces.