A Real-Time Method for Detecting Temporary Process Variants in Event Log Data

During the execution of a business process, organizations or individual employees may introduce mistakes, as well as temporary or permanent changes to the process. Such mistakes and changes in the process can introduce anomalies and deviations in the event logs, which in turn introduce temporary and periodic process variants. Early identification of such deviations from the most common types of cases can help an organization to act on them. Keeping this problem in focus, we developed a method that can discover temporary and periodic changes to processes in event log data in real-time. The method classifies cases into common, periodic, temporary, and anomalous cases. The proposed method is evaluated using synthetic and real-world data with promising results.

Automated summary

During business process execution, mistakes and changes introduced by organizations or employees can lead to anomalies in event logs, which create temporary and periodic process variants. An early identification of these deviations from common cases can help organizations take action, and a method has been developed to classify cases into different categories for real-time discovery of process changes in event log data. The method was evaluated using synthetic and real-world data with promising results.