Fuzzy multi-perspective conformance checking for business processes

Conformance checking techniques are widely used to monitor the execution of organization processes and to pinpoint possible violations of the prescribed behavior. State-of-the-art approaches adopt a crisp evaluation of deviations: namely, every step in the execution which is not perfectly compliant with the procedural rules is marked as deviant. However, many real-world processes are driven by decisions taken by human actors, which are often characterized by uncertainty. As a consequence, deviations are often tolerated, within some boundaries. In these contexts, assessing small violations at the same level as significant ones hampers the accuracy of the provided diagnostics. In this work, we propose a novel conformance checking approach which allows to consider actors’ tolerance to violations when assessing the magnitude of detected deviations, taking into account different kinds of deviating behaviors. Experiments conducted on two real-life clinical data sets have shown that taking the extent of deviations into account leads to more fine-grained diagnostics, thus illustrating the value of the approach.