Computing Reviews

Evaluating The Mediator Method:Prism as a Case Study
Sullivan K., Kalet I., Notkin D. IEEE Transactions on Software Engineering22(8):563-579,1996.Type:Article
Date Reviewed: 07/01/97

The authors present a case study of their experience in applying the mediator method in the development of a complex software environment with highly integrated behavioral relationships. The mediator method, described in a previous paper by Sullivan and Notkin [1], integrates functional components of a design through the use of well-defined interface descriptions and the use of mediator agents to describe how functional components work together. A directed graph, whose nodes represent component behaviors and whose edges represent behavior relationships, provides the basis for developing functional abstract behavior types. In the mediator approach, abstract behavior types are the fundamental  design  element; they are basically abstract data types that can export their associated events.

The object of the case study is the development of a suite of software tools, collectively known as the Prism system, that are used to plan radiation treatments for cancer patients. The elements of the Prism system are highly interrelated and include interactive management of images and iterative development of proposed treatment plans. In such a highly integrated environment, maintaining consistency among the tools is critical to the efficient and accurate operation of the system. The paper explores many of these critical relationships in detail and shows how specific capabilities of the mediator method facilitated the development of the Prism system. The success of this design project proves that the mediator method is a viable and capable approach. The behavior-oriented modeling techniques underpinning the method are not intuitive, however, and can be obtained only from practical experience and a significant change in the traditional design perspective.


1)

Sullivan, K. J. and Notkin, D. Reconciling environment integration and software evolution. ACM Trans. Softw. Eng. Method. 1, 3 (July 1992), 229–268.

Reviewer:  W. T. Neumann Review #: CR120597 (9707-0523)

Reproduction in whole or in part without permission is prohibited.   Copyright 2024 ComputingReviews.com™
Terms of Use
| Privacy Policy