Managing domain architecture evolution through adaptive use case and business rule models
Carl Robert Carlson, Russell R. Hurlbut · 1998
The areas of domain engineering, vertical application frameworks, and business objects have generated considerable interest in industry and the research community during the last few years. In order for systems to be successfully implemented from such application frameworks, there are two major concerns that must be addressed: how to maintain non-interfering applications and how to minimize bias towards any individual application in developing the domain model. As we add new applications, we need to ensure that they are not destructive to each other. We also need to make certain that as we expand the domain model to accommodate new applications, that these future applications are not unnecessarily constrained or complex. Most of the related work to date has focused on defining the architectural structure of application frameworks or on maintaining structural and behavioral consistency. Methodologies have primarily focused on the definition of a domain model rather than its evolution. This thesis develops a conceptual framework for integrating various techniques to facilitate managing the evolution of a business domain architecture. As part of this conceptual framework, domain normal forms and normalization operations are defined. An adaptive use case model is proposed as an extension to the Unified Modeling Language (UML) specification. A business rule pattern language and meta-model are also developed that describes how parameterized business rules can be integrated with adaptive use cases to manage domain model evolution. All of these models are synthesized into a domain evolution architectural transformation (DEAT) process model. Although considerable work remains, this thesis demonstrates the feasibility of utilizing the prescribed models for guiding performance of operations on a domain model. Through fit assessment and change cost analysis, new applications may be developed from the domain architecture with minimal bias and interference resulting in a more stabile and resilient domain model.