A methodology for interschema relationship identification in heterogeneous databases

Information sharing among databases, whether scientific or business, requires the development of techniques for accessing data from multiple heterogeneous databases. One approach to providing interoperability among these databases is to define one or more federated schemas which represent a coherent view of the underlying databases. A review of existing research on schema integration (the process of generating integrated schemas) points to the need for the development of techniques for identifying objects in multiple databases that may be related. This process of interschema relationship identification (IRI) is the focus of this paper. We present a methodology for utilizing schematic as well as integrity constraint knowledge for IRI. We present the details of a comprehensive set of heuristics for schematic IRI. We also introduce the concept of constraint-based relationships and describe how such relationships can be generated. Finally, we describe heuristics for generating "real world" interschema relationships based on the schematic and constraint-based relationships that are generated.