On the Applicability of Schema Integration Techniques to Database Interoperation

1 I n t r o d u c t i o n Interoperation among pre-existing, heterogeneous, and autonomous databases has been an important research topic in the last few years. Recently, the trend in database interoperability research is moving towards architectures for interoperation of databases on a scale that goes beyond the context of a single organisation, exploiting the communication facilities offered by world-wide networks. The canonical model used is often an object-oriented one [1]. It has been recognised that such an environment requires flexible and scalable architectures, where users of the component databases are provided with tools to establish importation of information from remote data sources [2,3]. Tightly-coupled approaches, where the schemata of all component databases are unified into a single global schema by a central modelling authori ty possessing a helicopter view of all component databases [4], are generally agreed to be infeasible in such situations, if only because the component databases may be quite diverse, and no-one can be expected to grasp all information available in the interoperation environment. Two main approaches towards a more loosely-coupled style of database interoperation can be distinguished. In the multidatabase approach [2], users are expected to define their information needs using a powerful query language with constructs for on-the-fly semantic reconciliation of heterogeneous data. It has been argued, however, that this puts an unacceptable burden on the user, to whom a single logical view of the interoperable databases is no longer presented.