Behavior-inductive data modeling for enterprise information systems

Abstract

Traditional database design tools commonly have a restriction that their users are assumed to have expertise in entity-relationship (ER) modeling. What we have found is that once an enterprise-wide business description is prepared, even a novice field worker is able to obtain an ER model with the assistance of a design tool which automatically extracts data objects from the description and which semiautomatically classifies them into entities or attributes. Traditional entity-oriented automated database design tools have another limitation that a bunch of attribute redundancies can be induced by concealing or omitting some meaningful relationships. To avoid the major negative habits of traditional approaches, our design tool treats relationships rather than entities as the focal point in database design. Our results with an option trading application have shown that, with just a few interactions, field workers can use our tool to generate an appropriate ER model.