Mining Behavioral Patterns for Conformance Diagnostics

Conformance checking is the field of process mining concerned with the monitoring and reporting of discrepancies between event logs and process models. An often overlooked issue is the entry barrier for conformance checking techniques. On the one hand, constraint-checking methods provide intuitive conformance diagnostics, yet require a significant manual effort and expertise from the users to elicit the corresponding constraints. On the other hand, procedurally-oriented techniques, e.g., alignments, provide low-level conformance results that require a significant interpretation effort from the end-user. Therefore, in this paper, we propose to combine the best of both worlds and present an automated method to generate conformance diagnostics in the form of higher-level behavioral patterns, derived from a procedural model. The approach is implemented as a standalone tool and evaluated against real-life datasets, where it is shown to explain nearly all deviations with good scalability.