An Holistic Evaluation of Federated SPARQL Query Engine
Nur Aini Rakhmawati · ISICO 2013 · 2013
Federated SPARQL Query Frameworks have been actively developed recently. Such a system consists of federated engi nes as a mediator and a group of SPARQL Endpoints. The mediator receives a query from a client and distributes it to relevant SPARQL Endpoints. As the core component of federation systems, existing evaluation methodologi es usually only test the federated engines regardless of the capacity of the SPARQL Endpoints. Most evaluations are chiefly concerned about the speed o f the engines response as the sole indicator for federation performance. Henc e, the development of federation systems focuses primarily on source se lection and join query optimization in the mediator. Several approaches me rge a set of sub-queries into one query to reduce the response time. However , SPARQL Endpoints have a restriction to accept a query request in a c ertain period of time and return the results per query request. We hence cond ucted an experiment to evaluate federation frameworks by measuring their r esponse time and data transfer between mediator and the SPARQL Endpoints. Additionally, we introduce a set of new metrics that are more suitab le to assess the performance of federation frameworks performance in a range of environments.