INTEROPERABILITY IN INFORMATION-SYSTEMS

Two operationally autonomous information systems interoperate if one system can make use of information contained in the other, or if they exchange transactions. A widely held assumption in this field is that the development of interoperability among information systems should preserve the design autonomy of the individual information systems in a strong sense. This paper demonstrates that preservation of strong design autonomy is not always possible owing to the necessity to resolve semantic hereogeneity among the systems. This necessary violation of autonomy calls into question the utility of research on general architectures and the possibility of general solutions. We suggest some themes in certain special cases where we believe feasible and useful computer science research can be done in this area, and also suggest some implications for the standard models of architectures for federated databases.