Database design with behavior and views using parameterized Petri nets

Abstract

We propose a method and a modeling tool for the design of relational databases. This conceptual design technique incorporates, from the earliest design stages, both database behavior and the existence of multiple user views of the database. It utilizes the structure and capabilities of parameterized Petri nets (PPNs) described in [2]. Petri nets are traditionally found useful for describing conditions and events that constitute a dynamic system. Parameterized Petri nets are (non-strictly) heirarchical or multi-level, and have a parameterization capability which is used to develop differing useful system views (PPNviews) of the relational database structure and dynamics.

Using Petri nets in general and PPNs in particular, users may create models that describe the flow of work or the transformations that take place in the system at an intuitive, comfortable level of detail. This produces one view of the system (PPNview). Different users with detailed knowledge about particular segments or aspects of the system will be able to design PPNviews showing those sections in great detail while leaving the less familiar segments less detailed. A set of normalized relations produced by traditional methods is the basis (parameterization descriptor) for a different PPNview. This PPNview depicts the transactions that occur in the system connected to the normalized relations that participate in the transaction.