An Ontology connected to several data repositories: query processing steps

The great expansion of communication networks has made available to users a huge number of heterogeneous and autonomous data repositories that present different structures/organizations, query languages and data semantics. In that context it is clear that new information retrieval techniques with a strategy that focuses on information content and semantics are needed. We propose to use domain specific Ontologies to capture the information content of such repositories whenever available. We describe such Ontologies using a system based on Description Logics. In this paper we present all the stages of the processing of a query formulated over an Ontology when the answer must be found in the underlying data repositories. Those stages make up a subpart of the global query processing strategy defined for a set of loosely-coupled Ontologies. We show first how the query is transformed into a semantically equivalent one and how inconsistent queries are detected. Then, we explain the test to verify if the query can be answered from the cache memory. Next, we present a set of heuristics used during the query decomposition process. Later on, we show how to optimize plans associated to subqueries that access the underlying data repositories and finally we illustrate how the answers retrieved from the repositories are correlated in order to generate the query