Workflow Technology is emerging as one of the fastest growing disciplines in information technology. Workflow technology essentially provides automated coordination of business activities and the flow of information through the enterprise. However, workflow modelling plays a critical role in the understanding, analysing and possibly reengineering of business processes. A well-defined workflow conceptual model leads to the development of an effective and reliable workflow application. It is essential to correctly and effectively capture the many aspects of workflow specification requirements before their deployment through workflow management systems, since anything short would compromise the goals of the business process.
In this thesis, we look at the different issues in capturing such requirements and propose a systematic layered modelling approach. We split the workflow specification requirements into five basic dimensions: structure, data, execution, temporal, and transactional. While we acknowledge the importance of all dimensions of workflow specification, we believe that the foundation of a process model lies in its structural specifications. This thesis is primarily focussed on the structural aspect of workflow models. We initially present a generic core workflow modelling language that is characterized by simplicity and compactness. This graphical language has four basic modelling objects: task, condition, synchronizer, and flow. Using the four modelling objects, we identify the following workflow modelling constructs: ordering, alternative, exclusive join, concurrency, synchronization, iteration, start/stop, nesting, and contingency.
It is possible to easily get into error situations while building large workflow specifications. We show how a structural specification may contain deadlock and lack of synchronization conflicts that could compromise the correct execution of workflows. In general, identification of such conflicts is a computationally complex problem and requires the development of effective algorithms specific for the target modelling language. A major contribution of this thesis is new verification techniques for satisfying well-defined correctness criteria in process models. We present a visual verification approach and algorithm that employs a set of graph reduction rules to identify structural conflicts in process models for the given workflow modelling language. We provide insights into the correctness and complexity of the reduction process. We show how the reduction algorithm may be used to count possible instance subgraphs of a process model. We also present an alternative verification algorithm based on split/join analysis.
Generally, a process model evolves through numerous changes during its lifetime to meet dynamic and changing business requirements. It is essential that such changes are introduced systematically and their impact is clearly understood. Process model transformation is a suitable approach for this purpose. Applying pre-defined transformation operations can ensure that the modified process conforms to a given class of constraints specified in the original model. We identify three classes of transformation principles - equivalent, imply, and subsume - to manage changes in process models. A simple algebraic notation for representing process graphs is also presented that can be used to reason about transformation operations.
We map the generic modelling language to an extensible and explicit process modelling language. We define extended modelling structures to capture advanced workflow requirements conveniently. We look at alternative ways for modelling iteration in workflow models. The impact of these extensions on verification capabilities is discussed and verification algorithms are extended.
The modelling and verification concepts introduced in this thesis have been applied as a foundation to the development of a workflows modelling and verification tool, FlowMake. FlowMake demonstrates how the research results from this thesis can be deployed in real world workflow applications.