Impact of Threshold Setting for Event Log Repair on Conformance Checking

Area of Science:

• Business Process Management
• Data Quality
• Process Mining

Background:

• Conformance checking compares event logs to process models to find discrepancies.
• Low-quality event logs with noise can hinder accurate conformance checking.
• Existing log repair methods require difficult-to-set thresholds.

Purpose of the Study:

• To investigate how low-quality event log repair methods affect conformance checking results.
• To understand the impact of repair thresholds on the accuracy of identifying process differences.

Main Methods:

• Experimental analysis of different low-quality event log repair techniques.
• Evaluating conformance checking outcomes with repaired versus unrepaired event logs.
• Varying repair thresholds to assess their influence on results.

Main Results:

• The effectiveness of log repair methods is contingent on the specific event log.
• The optimal repair threshold is influenced by the amount of noise present in the log.
• Repairing low-quality logs can alter conformance checking outcomes significantly.

Conclusions:

• The choice of threshold for repairing low-quality event logs is critical for accurate conformance checking.
• Process mining practitioners must carefully consider event log characteristics when applying repair methods.
• Further research is needed to develop adaptive threshold-setting mechanisms for log repair.