
225Scalable End-User Access to Big Data
can be more efcient, and hence an OBDA system should provide the means
to dene such procedures (e.g., by means of domain-specic plug-ins).
To conclude, an optimal system for query answering through query
rewriting in the context of Big Data must be approached in an integral way,
including modules that handle and optimize each of the aspects of the query
answering process, while trying to maximize the benets that are obtained
by separating reasoning with respect to the ontology vs. reasoning with
respect to the data. The resulting architecture of such a system may look
like the one ...