Materialization of OWL Ontologies from Relational Databases: A Practical Approach

PABLO R. FILLOTTRANI; SERGIO GÓMEZ

Computer Science ? CACIC 2019, Communications in Computer and Information Science

Springer

Año: 2020; p. 285 - 301

Providing both end-users and applications with a uniformway to query legacy databases through a high-level ontology that modelsboth the business logic and the underlying data sources is the main con-cern in Ontology-based Data Access (OBDA). Our goal in this research isproviding tools for performing OBDA with relational and non-relationaldata sources. Within the OBDA framework, in this work, we presenta prototype tool that can access an H2 database, allowing the user toexplicitly express mappings, and populating an ontology that can besaved for later querying. We report on the current functionality of ourtool, which includes creating, loading, saving a global ontology populatedwith a database or a CSV file. For the latter, we devised a language forspecifying the underlying schema of the CSV file. We argue that this lan-guage is better suited than current alternatives such as JSON. Also, thesystem allows the user to visually express mappings from the database tothe ontology and the ability to create databases for testing the behaviorof the system in the presence of increasing workloads. Our tests indicatethat the system can handle a moderate workload of tables of tens ofthousands of records but fails to handle tables of millions of records.