Combining Description Logics and object oriented models in an information integration framework

Abstract

We present an information integration system called SINTAGMA which supports the semantic integration of heterogeneous information sources using a meta data driven approach. The main idea of SINTAGMA is to build a so called Model Warehouse, containing several layers of integrated models connected by mappings. At the bottom of this hierarchy there are the models representing the actual information sources. Higher level models represent virtual databases which can be queried, as the mappings provide a precise description of how to populate these virtual sources using the concrete ones. The implementation of SINTAGMA uses constraints and logic programming, for example, the complex queries are translated into Prolog goals. This paper focuses on a recent development in SINTAGMA allowing the information expert to use Description Logic (DL) based ontologies in the development of high abstraction level conceptual models. Querying these models is performed using the Closed World Assumption as we argue that traditional Open World DL reasoning is less appropriate in the context of database oriented information integration environments.