Efficient Consistency Check Based on Perceived Initial Deviation

The behavior recorded by an information system is often different from the behavior in the initial model since the business process is constantly changing in actual operation. In order to enable event logs to be replayed well to process the model, the set of optimal alignments first needs to be determined. All the corresponding activities in each alignment are compared to the existing method, which will cause a lot of unnecessary work. Thus, we propose that the non-optimal alignment is perceived beforehand according to the relationship between the location of the initial deviation and the number of deviations. The perceptible regions in the process model are divided based on the behavioral characteristics of various substructures. The comparison of an alignment is terminated if the location of the initial deviation is less than the previous value in the perceptible region. This alignment is judged to be non-optimal. Otherwise, the alignment with optimal probability is completely compared. The OPS plug-in was executed in the data sets from various networks and BPIC2020, and the results showed that the search efficiency could be improved under the premise of guaranteeing optimality.