Design Rule Hierarchy and Analytical Decision Model Transformation

We present two contributions to help make design rule theory operable in software engineering practice. First, we develop an approach to automatically transform a design, expressed in the prevailing unified modeling language, into an augmented constraint network (ACN), from which a design structure matrix (DSM) can be generated. Using ACNs formalizes design rule theory and using DSMS enables option-based reasoning. Second, we design an algorithm to derive a design rule hierarchy from an ACN, revealing the impact scope of each design rule and identifying the independent modules. This hierarchy defines an order in which decisions can be made to maximize task parallelism, constrained by the dependency structure of the architecture. We evaluate the accuracy and scalability of our approaches using both small, but canonical, systems and the open-source Apache Ant system.