Supervisor(s): prof.dr.ir. P.W.P.J. Grefen and dr.ir. R.M. Dijkman
Chair: prof.dr. I.E.J. Heynderickx
During the execution of an inter-organizational business-to-business (B2B) collaboration, a collaborating party may drop out for technical reasons or for business reasons. In such a case, the leaving party must be replaced, at runtime, by a new party. Ideally, the new party can pick up where the old party left off.
Currently, algorithms exist that can help with the runtime selection of the new party in order to incorporate it in the collaboration. Also, several studies have investigated the theoretical foundations of dynamic business process changes within a single enterprise. However, very few attempts have been made which can help a new party in a collaboration to pick up where the old party left off. Designing such techniques constitutes a challenge due to each party’s autonomy and to privacy policies that emerge in the context of a collaboration.
This PhD study aims to address this challenge by providing an overview of the components, algorithms, operations and techniques that are necessary to enable a party in a collaboration to be replaced by another party at runtime. Accordingly, this study consists of the following three research activities.
Firstly, it presents a descriptive reference architecture for Business Process Management Systems (BPMS) that facilitates switching parties in inter-organizational collaborations. This reference architecture, called BPMS-RA, has been designed based on a systematic literature survey of existing BPMS architectures. The main purpose of the development of BPMS-RA is twofold: (i) it can be employed as an architectural template for developing a new BPMS by offering two distinct levels of aggregation for the components that comprise a BPMS architecture, and (ii) it enables the analysis and comparison of existing BPMS in terms of their functionalities.
Secondly, it introduces a strategy for adapting an inter-organizational collaboration when a global view on the collaboration exists. In this strategy we assume that it is possible to have a central party, called the global controller, that can observe all communication and be the intermediary between all collaborating parties when one party is replaced by another. In this case there is also a model that describes all communications, which is called the choreography model. This research contributes to the body of knowledge by proposing a set of algorithms, operations and techniques (such as rollback and compensation) that facilitate the party switch in the case where a global controller and a choreography model exist.
Finally, it describes a strategy for adapting an inter-organizational collaboration when a global view on the collaboration does not exist. In this strategy, we assume that there is no global controller in a collaboration and no choreography model of the collaboration. The main challenge in this case is to capture the choreography model by relating the past communications among the collaborating parties that belong to the same thread of collaboration. This challenge is also known as the correlation challenge. This study addresses this challenge by introducing a new process discovery algorithm, called the correlation miner, that facilitates discovery when events (i.e., messages) are not associated with a case identifier.
In order to demonstrate the feasibility of these studies, we have developed prototype tools that implement our solutions and we evaluated them in a practical setting. Additional experiments were performed on both synthetic and real-world process models in order to determine the extent to which our proposed solutions are applicable.