OC\(\pi \): Object-Centric Process Insights

Process mining uses event sequences recorded in information systems to discover and analyze the process models that generated them. Traditional process mining techniques make two assumptions that often do not find correspondence in real-life event data: First, each event sequence is assumed to be of the same type, i.e., all sequences describe an instantiation of the same process. Second, events are assumed to exclusively belong to one sequence, i.e., not being shared between different sequences. In reality, these assumptions often do not hold. Events may be shared between multiple event sequences identified by objects, and these objects may be of different types describing different subprocesses. Assuming “unshared” events and homogeneously typed objects leads to misleading insights and neglects the opportunity of discovering insights about the interplay between different objects and object types. Object-centric process mining is the term for techniques addressing this more general problem setting of deriving process insights for event data with multiple objects. In this paper, we introduce the tool OC\(\pi \). OC\(\pi \) aims to make the process behind object-centric event data transparent to the user. It does so in two ways: First, we show frequent process executions, defined and visualized as a set of event sequences of different types that share events. The frequency is determined with respect to the activity attribute, i.e., these are object-centric variants. Second, we allow the user to filter infrequent executions and activities, discovering a mainstream process model in the form of an object-centric Petri net. Our tool is freely available for download (http://​ocpi.​ai/​).