ReLIGn:a Tool for Model Repair based on Local Instance Graphs

Model repair techniques update process models to incorporate behaviors observed in event logs, but not compliant with the original model. While these techniques address important practical needs, most state-of-the-art approaches repair anomalous behaviors independently, neglecting potential correlations among anomalies occurring at different process stages. This limitation introduces potential issues, e.g., over-permissive and low-quality models. In this paper, we present ReLIGn, a novel tool for process model repair that includes in the original model a high-level anomalous behavior (AB) represented as a Local Instance Graph (LIG). The tool supports the user in the evaluation of the repaired model, both graphically by highlighting the repaired part of the model and numerically by reporting the difference in terms of fitness, precision, generalization and simplicity between the original and repaired model.