Primitive‐Composite Approach for Structural Data Modeling

Abstract

Accurate, efficient data exchange is vital to improving productivity in the architecture‐engineering‐construction industry. Much of the current work in data exchange is focused on the development of data exchange standards. In examining the data models in these standards and other efforts, we have identified a number of problem areas, including overuse of aggregation hierarchies, nonhomogeneous characterization hierarchies, and replacing relationships with subclasses. To support easy exchange of data between engineering computer applications while avoiding the data modeling pitfalls, we have developed the primitive‐composite (P‐C) approach. The P‐C approach provides a methodology for defining the primitive data concepts of a domain as separate objects and for assembling those concepts to form the composite abstractions that engineers customarily use. By relating composite objects from multiple programs to the primitive concepts, the data in those programs can be shared as primitive concepts derived from one program's composite object and reassembled as another program's composite object without having to provide a direct mapping between the two programs. The paper presents a formal definition of the primitive‐composite approach and explores its application in structural engineering.