“Some methodology and representation problems for the semantics of prosaic application domains”

Abstract

As the primary targets for exercises of re-engineering and interoperability, so-called prosaic application domains and their associated systems provide interesting and practically useful case studies for a treatment of semantics from a methodological as well as cognitive point of view. (In this paper we treat cognition as the process of linking knowledge to perception inside an intelligent agent.)

In this context, semantics can be given a concrete interpretation in the form of constraints and rules on a data model. This allows among other things for an incremental definition of the meaning of conceptual schemas and specifications of information systems. The same treatment of semantics is relevant for re-engineering, reverse engineering and the establishment of interoperation since all these problems involve the reconstruction of lost knowledge when the “cognitive computation” was thrown away during the acquisition phase. No comprehensive or adequate methodologies seem to exist at this point that support this cognitive process in such a way that the computation can be recovered. We have briefly mentioned a few techniques that are relevant to the issue, such as induction and schema tranformation. A fundamental aspect of this problem area is that all semantic processes occur in heterogeneous groups, who must agree formally on specifications that in general will serve as input to CASE tools. Refinements of existing and well-tried methods such as NIAM are necessary; in general system methodologies will need to evolve in order to accommodate more linguistic-type knowledge generated during the specification acquisition process.