Workflow

Business process architecture and the Workflow Reference Model
© Chris Lawrence 2006

INTRODUCTION
This paper is a response to David Hollingsworth’s ‘The Workflow Reference Model 10 Years On’.1 He refers to a lack of ‘agreement on a common metamethodology for modeling the business process’. That is what this paper offers: a process meta-model. The meta-model is derived not by abstractingbusiness process architecture from workflow or Business Process Management (BPM) implementation technology, but by analysing ‘the business process in terms of its core characteristics’.2 Business rules play an important part in the meta-model, and the relationship between process and rule is compared with that of Ronald Ross’s ‘Business Rule Approach’. Implications of the meta-model are explored tosee to what extent they illuminate some of the process issues Hollingsworth raises in his paper.

BUSINESS PROCESS ARCHITECTURE
Hollingsworth claims the Workflow Reference Model (hereafter abbreviated to ‘WfRM’) was an attempt to construct an abstract view of the business process in terms of its core characteristics, separated from the technologies that could be used to deliver itsfunctionality in a real world situation.3 I am not sure it has done quite this. What it might have done instead is abstract general principles from workflow technology applications, in pursuit of effective interoperability standards for the industry. But this is different from constructing an ‘abstract view of the business process in terms of its core characteristics’. The latter would call for a conceptualmeta-model articulating what it is to be a business process, not just what it is to be a workflow system. Neither the WfRM nor the equivalent Business Process Management (BPM) reference model Hollingsworth proposes provides an analysis of what it is to be a business process. They both seem to take it for granted that we know what a business process is.
David Hollingsworth: ‘The WorkflowReference Model 10 Years On’, included in the WfMC Handbook 2004, Future Strategies Inc, Lighthouse Point, FL, 2004. 2 David Hollingsworth: ibid. 3 David Hollingsworth: ibid.
1

This paper offers something which I think is more like ‘an abstract view of the business process’, and one which in particular seems to fit the kind of rule-intensive contexts where workflow technology predominates, like‘insurance, banking, legal and general administration’ as well as ‘some classes of industrial and manufacturing applications’.4 I would argue that the WfRM is an abstraction from workflow technology, not a conceptual analysis of the process/workflow space in business terms. Either that problem space was deliberately seen as out of scope – which is fine; or it was assumed that abstracting from workflowtechnology offerings is the same as conceptually analysing the process/workflow problem space – which is not so fine. By extension the BPM reference model could end up as an abstraction from BPM technology offerings rather than the conceptual analysis of the process space in business terms which it needs to be. Hollingsworth’s paper5 displays other clues to what seems a generally (deliberately?)un-analytical approach to ‘process’ – as if assuming that conceptual analysis may not get us far. ‘Process fragment’ is a case in point: …more emphasis is required on the decomposition of processes into fragments and their consolidation in various ways to support more dynamic operational business processes. This stems from the vast increase in co-operating e-businesses brought about through theWorldwide Web… The original model identified various ways in which process fragments could interact – hierarchic subprocess, parallel synchronised processes, etc and did develop runtime models for binding them in execution terms. However, it did not attempt to develop anything beyond a primitive choreography capability in the area of process definition support for interactions between process...