Workflow Modeling Technologies

Introduction

Workflows, that is, well-defined sequences of tasks coordinated in order to achieve a variety of business, scientific and engineering goals, have emerged as a prominent technology in current distributed and dynamic environments, fuelled to a large extent by the development of Service Oriented Architectures (SOA) and their loose-coupling nature. Emanating from the first office automation systems, workflows originally had a purely business orientation and have in the meantime evolved to what is being referred to today as business workflow or, more broadly speaking, Business Process Management (BPM) technology. Indeed, the Workflow Management Coalition (WfMC) defined a workflow as:

However, it later became apparent that the workflow paradigm could also benefit the sciences domain and their complex and data-intensive operations. This has led to the emergence of a new family of workflows referred to as scientific; in general terms, a scientific workflow is:

Business and scientific workflows present similarities but also differences, mainly stemming from the purposes they serve and their historical context. In any case, a fundamental common characteristic is that all workflows rely on models, providing their conceptual representation, or, in other words, the “blueprint” from which the eventually executed workflow instances are derived.

In this context, the goal of this chapter is to comprehensively present the most influential technologies in the area of both business and scientific workflow modeling. The next section provides a brief overview on their evolution and associated key concepts, followed by three sections constituting an overview of the currently most prominent approaches. Before concluding, the chapter highlights main future research directions.

Background

Most researchers agree that workflow management and modeling technologies have their roots in the office automation systems that emerged in the 1970s (Ellis & Nutt, 1980). At those times, variants of Petri Nets (Petri, 1962) have been used in order to model related procedures. However, it took two more decades before they came to the spotlight.

In 1993, the Workflow Management Coalition (WfMC) was founded, and two years later the Reference Model (WfMC, 1995) was published, describing the major components and functions involved in a workflow’s lifecycle. It was followed, three years later, by the first specification of Workflow Process Definition Language (WPDL), which evolved to the contemporary XML Process Definition Language (XPDL) (WfMC, 2012). In the meantime, various languages emerged, often focusing on different aspects. Among them, the Business Process Model and Notation (BPMN) (OMG, 2011) is the de facto modeling standard, whereas Yet Another Workflow Language (YAWL) (van der Aalst & ter Hofstede, 2005) is a particularly noteworthy approach coming from academia.

At this point, the use of the term workflow should be clarified, especially against the term business process. According to the WfMC, a business process is related to any kind of activity, manual or automated, that realizes a business objective. A workflow, on the other hand, is an (partial) automation of a business process. Following this distinction, and as explained in (van der Aalst, ter Hofstede & Weske, 2003), Workflow Management (WFM) focuses on creation and enactment of operational processes, whereas the more recent term of Business Process Management (BPM) constitutes a superset of the traditional WFM approach, extending it by support for other important aspects, as are Business Process Analysis (BPA), but also new ways to support operational processes.