Jolita Ralyté, Colette Rolland
Centre de Recherche en Informatique Université Paris 1 Sorbonne 90, rue de Tolbiac, 75013 Paris, France e-mail : ralyte, email@example.com
The need for a better productivity of system engineering teams, as well as a better quality of products motivates the development of solutions to adapt methods to the project situation at hand. This is known as situational method engineering. In this paper we propose a generic process model to support the construction of a new method by assembling method chunks generated from different methods that are stored in a method base. The emphasis is on the guidance provided by the process model, as well as on the means underlying guidelines such as similarity measures and assembly operators. The process model is exemplified with a case study.
We are concerned with Situational Method Engineering (SME). SME aims at defining information systems development methods by reusing and assembling different existing method fragments. The term method fragment was coined by Harmsen in [Harmsen 94] by analogy with the notion of a software component. Similarly to the component driven construction of software systems, SME promotes the construction of a method by assembling reusable method fragments stored in some method base [Seaki 93], [Harmsen 97], [Rolland 98a], [Ralyte 99b]. As a consequence SME, favours the construction of modular methods that can be modified and augmented to meet the requirements of a given situation [Harmsen 94], [Slooten 93]. Therefore, a method is viewed as a collection of method fragments that we prefer to call method chunks [Rolland 96], [Ralyte 99b] to emphasise the coherency and autonomy of such method modules. New methods can be constructed by selecting fragments/chunks from different methods which are the most appropriate to a given situation [Brinkkemper 98], [Plihon 98]. Thus, method fragments/chunks are the basic building blocks which allow to construct methods in a modular way. The objective of our work is to propose a complete approach for method engineering based on a method chunk assembly technique. In previous papers [Rolland 98a], [Ralyte 99b] we presented a modular method meta-model allowing to represent any method as an assembly of the reusable method chunks. In this paper we are dealing with the method chunk assembly process. We present a generic process model, the Assembly Process Model (APM), to guide the assembly of method chunks using different strategies depending on the type of situation in which the assembly activity has to be carried out. Chunk assembly is the support of situational method engineering and therefore we propose a Method Engineering Process Model (MEPM) providing several different ways to assemble chunks with the objective of constructing new methods or enhancing the existing methods by new models and/or new ways of working. Whereas the APM views the assembly of method chunks ‘in the small’, the MCPM takes a broader view where assembling method chunks is part of a larger method engineering process. As a consequence, the APM is embedded in the MEPM.
Both process models, namely the APM and the MEPM, are expressed using the same notations provided by a process meta-model. A process meta-model is an abstraction of different process models, i.e. a process model is an instance of a process meta-model. In this paper, we use the strategic process meta-model presented in [Rolland 99] and [Benjamen 99]. Following this meta-model, a process model is presented as a map and a set of associated guidelines. Such representation of the process model allows us to provide a strategic view of different processes. Indeed, this view tells what can be achieved - the intention, and which strategy can be employed to achieve it. We separate the strategic aspect from the tactical aspect by representing the former in the method map and...