Design of enterprise information systems is a problem-solving activity. A system architect, designer and programmer make numerous decisions about the structure and behaviour of the system on various levels. These decisions define the quality of the system under design (SuD) in all its aspects. An example of an application-level decision is whether to structure the domain logic according to a domain model, a table module or a transaction script. We want to investigate the effects of such decisions on quality attributes of software. This will allow us to make better software and to predict the quality of software before it is built. In this research, we try to empirically validate or reject hypotheses like: ¿In the majority of systems above 500 function points, systems with a domain model have better changeability than systems with a table module.¿ If the validity of such hypotheses depend on the context of the system, we want to know in which cases the hypotheses hold and in which they do not. To be able to do such empirical research, we first need to develop a theoretical framework that defines the research context. This framework defines concepts like design problems, options and quality indicators. The design problems and options define choices a systems designer makes when designing a system. The quality indicators define if an option is better than another option: the notion of ¿better¿ is operationalized by means of quality indicators. The three together form the design space. Other design space models are discussed in section 4. The goal of this paper is to present a design space as a framework for empirical research.
|Name||CTIT Technical Report Series|