The NPD Benchmark: Reality Check for OBDA Systems

In the last decades we moved from a world in which an enterprise had one central database—rather small for todays’ standards—to a world in which many different—and big—databases must interact and operate, providing the user an integrated and understandable view of the data. Ontology-Based Data Access (OBDA) is becoming a popular approach to cope with this new scenario. OBDA separates the user from the data sources by means of a conceptual view of the data (ontology) that provides clients with a convenient query vocabulary. The ontology is connected to the data sources through a declarative specification given in terms of mappings. Although prototype OBDA systems providing the ability to answer SPARQL queries over the ontology are available, a significant challenge remains when it comes to use these systems in industrial environments: performance. To properly evaluate OBDA systems, benchmarks tailored towards the requirements in this setting are needed. In this work, we propose a novel benchmark for OBDA systems based on real data coming from the oil industry: the Norwegian Petroleum Directorate (NPD) FactPages. Our benchmark comes with novel techniques to generate, from the NPD data, datasets of increasing size, taking into account the requirements dictated by the OBDA setting. We validate our benchmark on significant OBDA systems, showing that it is more adequate than previous benchmarks not tailored for OBDA.