The thesis is divided into seven chapters (cf. Figure 1-3). After this introduction and problem formulation, Chapter 2 surveys major lines of research on ISD methods, defines ISD, and characterizes the role of a method and tools in its enactment. We shall also clarify mutual relationships between ISD methods and computer-aided environments such as CASE tools. In Chapter 3 we survey the literature on metamodeling and method engineering. Our goal here is first to introduce some principle elements of ME, and second to examine what kind of ME tasks and metamodeling languages have been proposed to address method development. The limitations of current ME approaches, especially related to representing method knowledge and improving methods in use, form a motivation for the development of an incremental ME approach.

FIGURE 1-3 Structure of the thesis.

In Chapter 4 we study the issue of representing method knowledge, modeling and implementing a large portion of ISD methods into a modeling tool. Thus, our focus here is on studying the first research question, “how completely can meta-data models represent knowledge about ISD methods for modeling tools?”. We study what constructs are needed for metamodeling languages by using content analysis to obtain method knowledge from 17 ISD methods. This leads us to propose some extensions to ME and especially to the languages it applies.

Chapters 5 and 6 concentrate on the second research question: “how can experience of method use together with metamodels be applied for method refinements?”. Chapter 5 puts forward principles of incremental ME, and studies method development through experience-based method refinement. We propose some ideas for understanding method evolution and the dynamic nature of ME: how the applicability of methods can be evaluated and maintained in changing ISD environments. These principles are applied in two cases (Chapter 6) in which we study local method development in practice. We apply the proposed ME principles together with existing tools to investigate the incremental ME approach. In these studies our primary interest is not on how efficiently an organization develops ISs, but on how they learn about methods during their use, and how this knowledge can be incorporated back into methods. Finally, in Chapter 7 we shall recapitulate major findings of the thesis, and propose issues inviting future research.