Abstract
In Description Logics (DLs) and in the ontology-based data access (OBDA) scenario, the use of actual datatypes (such as those adopted in DBMSs) has received only limited attention, although datatypes, with their predefined semantics, might have a substantial impact on the computational complexity of inference in ontology systems. In this work we aim at overcoming these limitations, and study the impact of adding datatypes to the OBDA scenario. To this aim, we introduce a language for datatypes and we define the notion of a datatype lattice, constituted by a set of datatypes that depend on each other. We classify lattices based into a hierarchy of three classes according to distinguishing properties. and we establish a theoretical framework for datatypes in the OBDA scenario, based on three major components (a DL, a class of datatype lattices, and a query language). We establish the computational complexity of query answering for various significant instantiations of this framework, as ranging from FOL-rewritable to coNP in data complexity.