Behavior-inductive data modeling for enterprise information systems

Abstract

The absence of a dedicated module that can extract entities from business requirements has hindered practical use of traditional Automated Database Design (ADD for short) systems in business data modeling. Because a separated list of entities is required for the initial input, field workers could get little support from a traditional system unless they are well acquainted with an Entity-Relationship Diagram (ERD for short). Over the last few decades, considerable efforts have been directed toward a knowledge-based system for the entity extraction. However, due to excessive reliance on stale knowledge of past object classifications, most traditional knowledge-based ADD systems have failed to formulate an appropriate ERD for up-to-date business requirements. Wondering if there would be any other way to perform data modeling without reliance on prior knowledge is a major motivation for this study. We propose a new modeling method which can formulate a flexible ERD on the basis of business descriptions instead of prior knowledge. Rather than treating data objects as the focal point in database design, our method focuses on core behaviors in the business descriptions. To evaluate the performance of the proposed method, we developed a new design system and conducted a case study on option trading applications with the system.