Customizing Activity Behaviour for Flexible Business Process Execution

Business process management (BPM) technology plays a major role in current enterprise-wide information systems. In general, BPM aims to provide overall business process control and monitoring in order to increase the productivity of enterprise wide functions. By using methods, techniques, and software to design, enact, control and analyse operational processes, BPM facilitates the interplay of people and organisation. Despite its success in a variety of industries, several limitations and challenges have emerged from BPM solutions. These include the flexibility of the modelling languages, measurement of process effectiveness and process efficiency, dynamic adaptation to external changes, and consideration of data dependency thus disconnecting the process models from overarching business semantics as well as underlying IT systems. This thesis is motivated by the requirement to improve a number of existing limitations of business process modelling, process execution and process control. We aim to identify the current limitations by a comprehensive review of the state-of the-art of BPM and introduce a well formalised extension to address the limitations. In particular, we found that the handling of process activities was highly inflexible and not indicative of work practices. We thereby investigate the characteristics of the process activity and propose to customize activity handling through new modelling and constraints. The new modelling extension relaxes the constraints to enable greater precision when modelling inter-task dependencies. Through the specification of a user defined finite state machines to represent each activity, the activity execution behaviour can be captured more accurately. To begin with, we demonstrate the applicability of an existing verification algorithm in ascertaining the structural integrity of the more descriptive process models. Next we examine the current workflow specification language in relation to the execution of activity by presenting a motivating business scenario showing the limitation by the granularity of activity. In addition to these we introduce a customisable activity finite state machine allowing for more precise representation of complex activity states, and propose a new constraint to specify partial activity completion. Furthermore, we propose a practically driven extension to the function of workflow management systems by offering the merging and grouping of work items. In addition, we identify the requirement of monitoring and tracking process execution. We will present our analysis and recommendation guidelines for users to select suitable data structure to store ids. Finally, we propose to synergise the process and data layers through the introduction of data dependency constraints. The contribution is a presentation of an implementation which automates the translation of constraints for enforcement in the data layer. We conclude the thesis with the direction of further extension of prototype implementation