Empirical validation of knowledge-based systems for conceptual database design

Conceptual database design is a complex and difficult task for non-expert designers. However, many of the commonly committed errors can be prevented with a knowledge-based (KB) design support system. The interface to such a system can be programmed using one of two strategies: (1) restrictive strategy in which the user is forced to follow a specific problem solving path or (2) the guidance strategy in which the user is advised on possible next steps in the problem solving process. This study involves the development of two versions of a KB system – one with a restrictive interface and other with a guidance interface – and a control system that offers no KB help. In a lab experiment non-expert designers solved a difficult data modeling task using one of the three systems. Analysis of their performance indicates that the KB systems improve the users’ model accuracy. However, there was no significant difference in performance between the two KB system implementations. Subjects in the restrictive interface group rated their system as easier to use than the guidance interface group users.