Interoperability by Exchanging Executable Content

In the area of geographical information systems (GIS), the requirements posed upon a set of heterogeneous but interoperable systems are not easily met. The particular systems most often possess a certain degree of autonomy and are designed to meet specific requirements for their respective application domains. In order to be more than mere data exchange, interoperability needs to be based on a framework which is easy to understand for all participating communities and yet expressive enough to meet those application-specific requirements. One of the most difficult problems when designing such a framework is to find an appropriate depth or thoroughness of the items specified, that is, the question of what should be specified and what should be left to particular implementations. An interoperability specification that leaves many things open to the implementation is usually easy applicable but maybe not very useful if important things have been left out. In contrast, a specification that includes too many details and assumptions might avoid semantic heterogeneities and similar problems, but fail in acceptance by the software industry because it leaves no room for innovation and competitive differences. The decision on how much to specify is basically a question of the level of abstraction and can be compared to various abstraction levels that are used, for example, in layered software architecture (e.g., the ISO/OSI communication model; Stevens 1990). Such architectures are a result of vertical (in terms of abstraction) and horizontal (in terms of purpose) decomposition of complex computing systems, where the borders represent the interfaces defining the decomposition.