Probabilistic Compliance in Declarative Process Mining

When addressing real-world processes, it is crucial to account for their intrinsic uncertainty to better reflect the nature of such processes. In this work, we introduce the concept of Probabilistic Declarative Process Specification (PDS), which is based on the Distribution Semantics from Probabilistic Logic Programming, in order to describe declarative process models with both crisp and probabilistic constraints. The probability associated to a constraint represents its strength or importance in a specific process domain. From this, we propose a novel notion of probabilistic compliance of a process trace w.r.t. a PDS, and how to compute it with an existing algorithm. Preliminary experimental results on a healthcare protocol are presented to evaluate the feasibility of our proposed semantics on process conformance checking.