A novel trace-based sampling method for conformance checking.

Area of Science:

• Data Science
• Business Process Management
• Process Mining

Background:

• Organizations need accurate business process execution and policy compliance.
• Process mining analyzes business process data for improvement.
• Conformance checking verifies real process behavior against models.

Purpose of the Study:

• To address computational challenges in conformance checking for large event logs.
• To develop a trace selection approach for efficient conformance analysis.
• To reduce processing time and computational cost without sacrificing confidence.

Main Methods:

• A novel conformance checking method using representative trace sampling.
• Incorporation of data dispersion using a statistical measure.
• Determination of sample size and trace selection criteria based on dispersion.

Main Results:

• The method effectively reduces processing time and computational cost.
• Validation using fitness, precision, generalization, and processing time metrics.
• Demonstrated effectiveness on real-world health and synthetic event logs.

Conclusions:

• The proposed method efficiently handles conformance checking in large event logs.
• Trace selection based on data dispersion offers a viable solution to big data challenges in process mining.
• The approach maintains confidence in conformance values while improving performance.