John Mylopoulos

From information system requirements to designs: a mapping framework

Abstract Comprehensive methodologies for information system development need to provide a framework for the adequate representation of system requirements and also for their usage in generating system designs. Requirements specifications are assumed to include a functional description of what the information system is intended to do, how it will interact with its environment, what information it will manage and how that information relates to the systems environment. p]The generation of a design is achieved by mapping elements of the requirements model into one or more corresponding design objects. This mapping process is guided by two considerations. Locally, the process is directed by dependency types among requirements and design objects which determine allowable mappings for a particular requirements object. Globally, the process is guided by non-functional requirements, such as accuracy and security requirements on the intended system, which are represented as goals describing desirable properties of the intended system. Satisficing methods for these goals are used to guide local mapping decisions. p]The paper includes the description of a prototype implementation—called IRIS—of aspects of the proposed mapping framework and illustrates its features through a sample session. The implementation was carried out within the DAIDA project at the Institute of Computer Science of the Foundation for Research and Technology, Crete.

The TaxisDL Software Description Language

The purpose of the TaxisDL language is to express the conceptual design of an information system. The focus of the design process includes the data classes of the proposed system, the functions and transactions manipulating them, and the larger conceptual groupings of these actions into long-term activities, which we call scripts. The design of the language is based on ideas from semantic data models and formal specification languages.

Properties of Information Modeling Techniques for Information Systems Engineering

ion mechanisms. These determine the proposed organization of an information base using a particular conceptual model. This is a fundamental concern for conceptual models because organizations that are natural and intuitive lead to more usable information bases which can be searched effectively and can grow without users losing track of their contents. Tools. If an information base is to scale up and remain useful for a long time, it needs tools which perform information base operations efficiently, also ones that support analysis of its contents, to give users confidence that they are correct and consistent. The reader may have noticed that the proposed characterization ignores the methodologies supported by a particular conceptual model. This omission is deliberate. All methodologies that have been proposed, including ones used in practice, are specific to particular uses one intends for an information base. For instance, using an information base for requirements engineering, e.g., [Coad89], calls for a very different methodology than, say, one used for data modeling [Batini92], or knowledge engineering in AI [Hayes-Roth83]. Properties of Information Modeling Techniques 31

Mapping Information System Requirements to Designs

Comprehensive methodologies for information system development need to provide a framework for the adequate representation of system requirements and also for their usage in generating system designs. The generation of a design is achieved by mapping elements of the requirements model into one or more corresponding design objects. This mapping process is guided by two considerations. Locally, the process is directed by dependency types among requirements and design objects which determine allowable mappings for a particular requirements object. Globally, the process is guided by non-functional requirements which are represented as goals describing desirable properties of the intended system. Satisficing methods for these goals are used to guide local mapping decisions. The chapter includes the description of a prototype implementation — called IRIS — of aspects of the proposed mapping framework and illustrates its features through a sample session.