Improving Documentation by Repairing Event Logs

In enterprises, business process models are used for capturing as-is business processes. During process enactment correct documentation is important to ensure quality, to support compliance analysis, and to allow for correct accounting. Missing documentation of performed activities can be directly translated into lost income, if accounting is based on documentation. Still, many processes are manually documented in enterprises. As a result, activities might be missing from the documentation, even though they were performed. In this paper, we make use of process knowledge captured in process models, and provide a method to repair missing entries in the logs. The repaired logs can be used for direct feedback to ensure correct documentation, i.e., participants can be asked to check, whether they forgot to document activities that should have happened according to the process models. We realize the repair by combining stochastic Petri nets, alignments, and Bayesian networks. We evaluate the results using both synthetic data and real event data from a Dutch hospital.