Abstract
Data integration systems provide access to a set of heterogeneous, autonomous data sources through a so-called global, or mediated view. There is a general consensus that the best way to describe the global view is through a conceptual data model, and that there are basically two approaches for designing a data integration system. In the global-as-view approach, one defines the concepts in the global schema as views over the sources, whereas in the local-as-view approach, one characterizes the sources as views over the global schema. It is well known that processing queries in the latter approach is similar to query answering with incomplete information, and, therefore, is a complex task. On the other hand, it is a common opinion that query processing is much easier in the former approach. In this paper we show the surprising result that, when the global schema is expressed in terms of a conceptual data model, even a very simple one, query processing becomes difficult in the global-as-view approach also. We demonstrate that the problem of incomplete information arises in this case too, and we illustrate some basic techniques for effectively answering queries posed to the global schema of the data integration system.