Abstract
Ontology-based data access (OBDA) is a well-established approach to information management which facilitates the access to a (single) relational data source through the mediation of a high-level ontology, and the use of a declarative mapping linking the data layer to the ontology. We formally introduce here the notion of ontology-based data federation (OBDF) to denote a framework that combines OBDA with a data federation layer where multiple, possibly heterogeneous sources are virtually exposed as a single relational database. We discuss opportunities and challenges of OBDF, and provide techniques to deliver efficient query answering in an OBDF setting. Such techniques are validated through an extensive experimental evaluation based on the Berlin SPARQL Benchmark.