J. Kelleher (South Africa) and M. Simonsson (Sweden)
UML 2.0, Requirement Metamodel, Requirement Patterns, Traceability 1. Context and Motivation The global economic slowdown has resulted in the reluctance of software companies to begin new developments. Many companies are opting for the cheaper option of updating and maintaining existing product-lines. [9] The development of families of complex software intensive products has become a reality. [25] This new complexity combined with tighter regulations on requirement compliance and produ
Changes to the UML 2.0 revision indicate that clarifications on the future of traditional use case needs to take place. The indications are that the use case notation can be replaced by use-case classes. Use case classes model requirement types as cohesive package consisting of requirement attributes and operations. The UML 2.0 supporting documentation does little to demystify the exploitation of the use case classes. The layered Model Driven Architecture and the underlying techniques recommended by the OMG may provide the solution. In this paper we review the critical changes to UML 2.0 regarding requirements modeling. We demonstrate that use case classes are formal templates for describing rules on modeling requirements with instances. We present a class hierarchical structure representing the complex relationship between business, product and project requirements using UML dependencies. We adapt the UML 2.0 extension mechanism notation to assist with the depiction of the requirement traceability links. Replacing use cases with classes and utilizing the explicit traceability links we integrate the requirement elements and the design elements into the same work space. Bridging the gap between requirements and the rest of the development process makes the test effort easier. For example, the use case classes serve as system and function test drivers. We show that requirement patterns are descriptions of communicating objects and classes. The UML extension mechanisms assist us to represent the requirement traceability links. Therefore by recognizing recurring dependency relationships of we are in fact uncovering traceability patterns. We support our project with a simple requirement process framework. Overall, we propose a new requirement modeling approach for the visualization, communication and reuse of technologies.
Important Links:
Go Back
