Temporal Conformance Checking at Runtime based on Time-infused Process Models
This work addresses runtime process monitoring for domains like finance and manufacturing, but it is incremental as it builds on existing multi-perspective mining approaches.
The paper tackles the problem of quantifying temporal deviations in conformance checking by infusing process models with a temporal profile based on task durations and event distances, and the evaluation on real-world financial and manufacturing datasets shows promise for improving runtime process monitoring and control.
Conformance checking quantifies the deviations between a set of traces in a given process log and a set of possible traces defined by a process model. Current approaches mostly focus on added or missing events. Lately, multi-perspective mining has provided means to check for conformance with time and resource constraints encoded as data elements. This paper presents an approach for quantifying temporal deviations in conformance checking based on infusing the input process model with a temporal profile. The temporal profile is calculated based on an associated process log considering task durations and the temporal distance between events. Moreover, a simple semantic annotation on tasks in the process model signifies their importance with respect to time. During runtime, deviations between an event stream and the process model with the temporal profile are quantified through a cost function for temporal deviations. The evaluation of the approach shows that the results for two real-world data sets from the financial and a manufacturing domain hold the promise to improve runtime process monitoring and control capabilities.