Checking Constraints for Object-Centric Process Executions

Conformance-checking techniques reveal the deviations between event data and the desired process specification, which can be expressed as a process model or a set of rules. State-of-the-art approaches assume a single case identifier, i.e., each case in the business process is associated with only one object. In contrast, processes in real life usually involve multiple object types. For instance, an order management process involves object types such as orders, items, and packages. These objects interact with one another, e.g., packing multiple items from the inventory to create a package. Existing techniques may provide misleading insights when applied to such object-centric event data. We address the issue by extracting process executions (cases) from the object-centric event log and representing constraints using Object-Centric Constraint Models (OCCMs). In this way, we handle cardinality, temporal, and performance constraints. Compared to procedural languages like Petri nets, the declarative nature of OCCMs provides more flexibility in modeling constraints, and constraint checking delivers more comprehensive diagnostics that go beyond isolated cases. The proposed method has been implemented as a ProM plug-in that supports the extraction of process executions, user-defined OCCMs, and constraint-checking. The feasibility of the proposed approach has been evaluated with other state-of-the-art approaches.