|
Process modelling is based on a process meta-model which follows
a situation and decision based approach.
The process meta-model comprises a reduced set of concepts (essentially
executable, choice and plan contexts) allowing to generate by instantiation
process models that we call ways-of-working. The major advantages of this
process meta-model are the following :
- It is a means for expressing heuristic knowledge useful for engineers
during the RE process
- It allows to generate ways-of-working at different levels of granularity,
one can describe large grained ways-of-working such as the complete development
of an application as well as very fine grained ways-of-working such as
the construction of an entity-type in an ER schema [Schmitt 95, Plihon
96].
- the generated ways-of-working are very modular and not monolithic
as in most approaches. A way-of-working is described as a set of independent
process chunks. This allow flexibility in the description as well as in
the evolution of ways-of-working.
- the process meta-model is fully generic, it allows to describe any
kind of decision based processes.
Concerning the process engineering aspects, we propose an enactment
mechanism called the "guidance engine" [Sisaid 96] and a meta-way-of-working
for guiding method engineers while constructing ways-of-working. The guidance
engine :
- allows to enact any way-of-working defined as instance of the process
meta-model. This enactment can be dynamic, ways-of-working can be changed
and enacted on the fly.
- is flexible because when something is proposed to the requirements
engineer by the tool, he has the ability to do something else : he can
choose another alternative than the ones proposed by a choice context,
he can abort a plan context at any time, etc..
- is fully generic, it enacts products and processes of different
nature at different level of granularity. It can also enact the meta-way-of-working,
i.e. the process model describing the way to construct process models of
any RE methodology.
The meta-way-of-working is based on the reuse of method knowledge using
a library of generic formulas. These formulas are a powerful means for
quickly generating process models. The meta-way-of-working is fully generic
in the sense that it can be used for the construction of any type of ways-of-working.
Finally, we demonstrate that the construction of ways-of-working as well as the process meta-model itself is based on a free algebra. This work offers a mathematical ground to our approach.
The process theory presented in this paper is currently used in two ESPRIT projects, namely CREWS (N° 21903) and ELEKTRA (N° 22927). Within the CREWS project, the theory is used to describe the various ways-of-working defining the life cycles of scenarios, scenarios being used for the sake of cooperative requirements engineering. In this project, following a situational approach, ways-of-working are also used to integrate different scenario based approaches in the overall RE process, i.e. to precisely characterise which, when, how and why scenarios should be used for the sake of RE. In the ELEKTRA project, the theory is applied for the sake of defining ways-of-working aiming at supporting the enterprise change process for re-organising electricity companies..
University of Aachen (RWTH-Aachen), Germany (Project Coordinator)
Universite Paris I (Sorbonne), Paris, France
City University, London, U.K.,
ICS-FORTH, University of Heraklion, Greece
SISU-ISE, Royal Technical University of Stockholm, Kista, Sweden.
[Plihon 96] : V. Plihon, " Un environnement pour l'ingénierie
des méthodes", Thèse de Doctorat, Université de
Paris I. Janvier 1996.
[Rolland 94] : Rolland C. : "A Contextual Approach to modeling the
Requirements Engineering Process", SEKE'94, 6th International Conference
on Software Engineering and Knowledge Engineering, Vilnius, Lithuania,
1994
[SiSaid96] S. Si-Said, C. Rolland, G. Grosz, "MENTOR : A Computer Aided Requirements Engineering Environment", in the Proceedings of the 8th CAISE Conference on Challenges In Modern Information Systems, Heraklion, Crete, Greece, May 1996
[Schmitt 95] : J.R. Schmitt, "Méta-modélisation des méthodes d'analyse des systèmes d'information", Thèse de Doctorat, Université de Paris VI. Décembre 1995.