Process interference: automated identification and repair

PhD ceremony: Mr. N.R.T.P. van Beest, 12.45 uur, Academiegebouw, Broerstraat 5, Groningen

Dissertation: Process interference: automated identification and repair

Promotor(s): prof. J.C. Wortmann

Faculty: Economics and Business

Concurrent execution of business processes is common in most organizations, though these processes may (partially) use the same data resources. Such mutual dependence on process variables may cause inconsistencies during process execution. Although these processes may properly terminate, they may lead to undesirable outcomes from a business perspective, due to changing data in concurrently running process instances. The situation where data is simultaneously modified by several processes is known as process interference.

In this thesis, the process interference problem is defined formally and analyzed at a large Energy company and a large Telecom company. Due to the complexity of the analysis, a software tool has been developed to analyze the processes and provide a complete overview of the erroneous situations. The analysis shows that process interference is far more common than a rare unfortunate exception.

In order to resolve this problem, dependency scopes are introduced to represent the dependencies between processes and data sources that are vulnerable to interference. Intervention processes are introduced to repair inconsistencies during execution of the process. Intervention processes are generated at runtime, by using domain-independent AI planning techniques.

An architecture supporting these concepts has been designed and implemented, in order to evaluate the feasibility of the approach. The results indicate that coupling dependency scopes with declarative goals and generating intervention processes at runtime by means of AI planning is a usable and realistic method for resolving erroneous path situations caused by process interference.