Computationally Feasible Query Answering over Spatio-thematic Ontologies

Providing query answering facilities at the conceptual level of a geographic data model requires deduction, and deduction in geographical information systems (GIS) is a demanding task due to the size of the data that are stored in secondary memory. In particular, this is the case for deductive query answering w.r.t. spatio-thematic ontologies, which provide a logical conceptualization of an application domain involving geographic data. For specific logics (socalled lightweight description logics) and query languages (conjunctive queries) the query answering problem can be solved by compiling the ontology-based query into an SQL query that is posed to the database. Thus, ontology-based query answering becomes as feasible as standard database query answering. In the literature, this kind of query answering by compilation is formalized using the notion of first-order logic (FOL) rewritability. In this paper we show that lightweight description logics such as DL-Lite can be combined with spatial calculi such as the region connection calculus such that FOL rewritability is retained and the expressive power is sufficient for modeling important aspects of GIS data. Keywords-description logics; qualitative spatial reasoning; deductive query answering; FOL rewritability