The problem of how to integrate multiple, disparate database systems into a single heterogeneous system is addressed. The authors’ approach is to define an object-oriented global schema, called a multi-information base (MIB). The MIB can contain explicit mappings to the individual databases resulting in a federated system, or it can contain views that integrate similar portions of the remote schemas into a single conceptual whole. Thus, a database designer has the flexibility to hide as much of the distributed nature of the database as he or she desires.

This paper cites two primary advantages of using an object-oriented model instead of the relational model. First, the mappings and view definitions can be more sophisticated, leading to better integration. Second, the remote systems can be of various data models, each having different interface languages.

This approach has been implemented in the KOPERNIK system. A significant portion of the paper is devoted to how schemas and mappings are specified in KOPERNIK, as well as how queries on the global schema are decomposed into local queries on the remote sites.