Abstraction and Context in Requirements Engineering: A Synthesis of Goal Refinement and Ethnography

Most requirements engineering (RE) research and practice embodies a philosophy that we will call abstractionism, which involves the building of simplified models of domains of discourse and proposed systems. Abstractionists make much use of formal models, such as goal dependency networks. An alternative design philosophy is contextualism, according to which the peculiarities of the context of use of a system must be understood in detail before the requirements can be derived. Contextualists use qualitative methods to uncover and help interpret these particularities. In this paper, we analyze what it would mean to combine the best features of abstractionism and contextualism, and we ground our discussion in an illustration of abstractionist and contextualist thinking about RE through goal refinement (GR). In the context of the domain of meeting scheduling, we contrast a wholly abstractionist approach to GR with one that incorporates data gathered using two ethnographic methods. In doing so, we consider each step of the abstractionist approach, illustrating where ethnographic data obtained in our work environment affects the model produced. As we proceed, we summarize the general lessons learned. We then discuss how other abstractionist and contextualist methods could be integrated.