The provision of software support for conceptual model design processes has been an important issue in Model Management Systems research. This study examines the issue by empirically studying design behaviour and the techniques used during design, with a focus on the use of abstraction in the design process.
A framework that explains the relationships between the use of abstraction and the output of the design process is proposed. We classify abstraction into vertical, horizontal and general abstraction techniques. We then propose that three dimensions of a design: the completeness of the design, higher level concepts in design, and the organization of the design, are affected by effective use of these abstractions. We also propose aids to support these three types of abstraction, and develop measures to evaluate their effectiveness.
A software prototype was developed to illustrate the implementation of the proposed abstraction aids. A pilot study was also conducted to test the effectiveness of these aids using different versions of the prototype.
Three hypotheses that test the effectiveness of the proposed abstraction aids were tested in an experimental study with treatments that included three design environments and two training methods. The three design environments were two which were supported by pencil and paper design, and one which was supported by a software prototype. The two training methods used were based on our proposed abstraction aids. The results of the experiment indicate some significant differences in the performance of the participants in different treatment groups.