Knowledge Component-based Architecture for Process Modelling Olivier Glassey Fraunhofer FOKUS, Kaiserin-Augusta-Allee 31, DE-10589 Berlin Tel +49 3463 7167, Fax +49 3463 8000 olivier.glassey@fokus.fraunhofer.de Abstract: In previous work we compared three process modelling techniques and applied them to e-Government processes. We identified major issues, amongst them the lack of formalism for knowledge modelling and of formal semantic links between different abstraction levels. In this paper we propose an architecture for process modelling that supports knowledge models and formally integrates them in process models. 1 Introduction Public administrations provide many different types of services to their clients, whether they are citizens, businesses, self-employed professionals, other units from the public sector, etc. The interactions between a public agency and its clients are mostly process-based and can be categorised as follow: structured procedures or routines, semi-structured decision processes and negotiation-based case-solving [Lenk & Traunmüller 1999]. Research in the field of business process modelling is quite active, also in the domain of e- government. The use of process models has many purposes. [Wyssusek & al. 2001] identified several of them in the literature and we selected those we found most relevant: facilitating human understanding, communication, organisational learning and transfer of know-how; supporting process improvement; benchmarking process performance; helping the integration of technical aspects and organisational issues. In previous work, the author and colleagues [Glassey & Chappelet 2002] studied two specialised process modelling methods: Adonis [Jungiger & al. 2000] and OSSAD [Chappelet & Snella 2004, Dumas & Charbonnel 1990] and furthermore used the UML graphical notation language [Booch & al. 1999] to create corresponding models. Although the main focus of UML is on information systems modelling, we were able to produce equivalent process models in most cases. This comparison aimed at providing a global view of an organisation and wanted to answer three general questions: – At the abstract level: what are the strategic goals of the organisation? – At the structural level: who is doing the work and which resources are available? – At the descriptive level: how does the organisation operate? We found this approach to be satisfying at the abstract and structural levels, but we discovered that it lacked several important concepts, particularly in terms of knowledge modelling. Indeed these process modelling techniques use concepts such as data, documents or information resources, but they do not take knowledge into account. We therefore tried to integrate a new layer in this general view of an organisation and introduced a new question: why is a given decision taken within a process? In this publication we describe how to integrate a knowledge component-based architecture into process modelling.