Discovery and Evaluation of Coordination Patterns for Business Processes in many-to-many Relationships

Today, organisations use process-oriented systems to manage and automate the enactment of their business processes. The cornerstone artifact is the process model, which at design-time is used to describe the steps that need to be fulfilled in order to reach a business goal. At run-time, the process model is executed and process instances are created. The existing modelling approaches are based on three main paradigms: the more traditional activity-centric paradigm, the case handling paradigm and the more recent data-centric paradigm.

Process models can be classified into monolithic and interacting process models. Monolithic process models are predominantly created in the activity-centric and case-handling paradigm. In a monolithic process model, all the involved resources and activities are contained in one vast model. In monolithic process models, interactions occur between the different partners involved in a cross-organisational setting which exchange messages with one another. Interacting process models are prevalent in the data-centric paradigm. In interacting process models, interdependent processes interact with one another such that on a meta-level a composite business process is achieved. In both types of models, interactions between interrelated processes need to be properly coordinated such that a common business objective can be reached.

Handling the complexity generated by highly interconnected scenarios, involving hundreds of processes, is a challenge in business process management. Process management systems for such collaborations must be capable of handling both synchronous and asynchronous process interactions. In the context of process management systems, different pattern catalogues such as the Service Interaction Pattern or Correlation Pattern have been used for describing fundamental types of interactions that repeatedly arise during business process modelling. Yet, until now, none of the existing pattern catalogues has explicitly tackled the interactions of heterogeneous business processes in a many-to-many relationship setting. Furthermore, the existing pattern catalogues for the interaction-perspective are not paradigm independent, but mainly focus on the activity-centric paradigm.

For modelling multiple interacting processes with different dependency constraints, a collection of patterns that explicitly describes interactions among processes in different types of relationships, in a paradigm-independent manner, is required. This thesis proposes a catalogue of patterns, named the Process Coordination Patterns, describing process interactions in a one-to-many and many-to-many relationship setting. In the developed pattern catalogue, the discovered seven patterns are illustrated by abstracting from any specific paradigm. The PCPs may be used as guidance for evaluating the degree to which existing approaches capture more complex process interactions. In this thesis, the proposed pattern catalogue is put into practice by evaluating the degree to which two modelling approaches, based on different paradigms, can support the seven Process Coordination Patterns.