Embedding data modelling in a general architecture for integrated information systems

Abstract

Current approaches for the development of information systems are too strongly biased towards implementation. The architecture developed here shows that a successful implementation depends primarily upon the correct analysis of the underlying conceptual structures and only secondarily on the right implementation choices. The data view is an example, where this separation is already successfully practiced, but function and organization views of current information systems design methodologies also need to incorporate these principles. However, also on the data view, more emphasis has to be laid upon the design and validation of conceptual models. Many problems are not resolved, especially if the models are large and involve interdisciplinary project teams. Naming issues, the identification of synonyms and of type conflicts still lack an efficent and systematic approach. Reference models, which are generic for a particular type of industry, can be used to improve the quality and reduce the development effort for enterprise-specific data models.[6] Their utilization should be incorporated into ISDMs and the corresponding CASE-tools. Furthermore, the architecture shows a clear need for integrating data models with function and organization models. This shall be the final step to truly automating information systems design.