RELATIONAL DEDUCTIVE OBJECT-ORIENTED MODELING (RDOOM) APPROACH FOR FINDING, REPRESENTING AND INTEGRATING APPLICATION-SPECIFIC CONCEPTS

Abstract

Next generation information systems (NGISs) have to support the manipulation of data-oriented, behavioral and deductive aspects of application domains. Many modeling methods, e.g. UML and other object-oriented modeling methods offer primitives for modeling data-oriented and behavioral aspects but they do not support the modeling of deductive aspects. In addition, NGISs may be implemented by several separate software/database systems that are based on different paradigms. Therefore it is not appropriate to use such a modeling method which is based on one paradigm. In NGISs it is essential to integrate the value-oriented approach with the object-oriented approach. We develop a relational deductive object-oriented modeling (RDOOM) approach for NGISs. Our goal is to combine into one modeling method, the navigation power of the relational model, the modeling power of the object-orientation and the inference power of the deductive (logical) framework. It is obvious that in NGISs complex and large specifications have to be embedded in application-specific concepts and structures that are defined beforehand for users to facilitate their query formulation. The detection and specification of application-specific concepts and structures means a new challenge for analysis methods. We show that on the basis of the primitives of RDOOM it is possible to represent this kind of application-specific information in a natural way. It is also necessary that the appropriateness and adequacy of application-specific concepts and structures can be tested before their expensive design and implementation phases. For the definition of these concepts and structures a diagrammatic representation typical of many modeling methods is not sufficient. Rather, a precise and executable representation is needed. Especially the complex and large derivation rules behind deductive concepts cannot be expressed precisely with diagrammatic representations. We develop set-theoretical representations for our primitives. The precise representation of the result of systems analysis also gives a more substantial starting point for the design and implementation of NGISs.