An empirical performance evaluation of a semantic-based data retrieving process from RDBs & RDF data storages

Abstract

SQL and, more recently, SPARQL are standard languages of the software industry for retrieving data. Other studies showed that retrieving data by using a SPARQL query is much slower than the semantically equivalent SQL query. Nevertheless, some recent proposals optimized database servers for SPARQL queries. This paper presents the results of a comparative analysis between the capability of SQL and SPARQL with respect to the retrieval of data from a relational database and RDF-triples. A free and open source-based scenario was constructed by using PostgreSQL and Virtuoso for storing data, and RETRI, a data retrieving software built in JavaScript which displays data views in a specific XML format. Open data from the British National Library Bibliographic Data Set were used in the experiment; results were analyzed from a performance perspective.