Efficient and effective business processes are highly relevant for any enterprise. As a consequence, the analysis and continuous optimization of business processes is a prerequisite for staying competitive. Usually, the execution of business processes is supported by process-aware information systems, which assist enterprise staff in executing the business processes. As opposed to domain specific software systems supporting individual processes, process-aware information systems may be reused in various scenarios to implement and execute various business processes. In other words, a process-aware information system can be applied in various domains (e.g., human resources, e-learning, or logistics). Consequently, the process model specifies the behavior supported by the information system, providing maximum guidance to users. In case a deviation from the modeled behavior is required, which has not been foreseen in the process model, the information system cannot support the necessary measures to fulfill the business needs.
Object-centric and data-driven business processes provide powerful means to tackle such problems and to increase flexibility. Instead of focusing on the control flow of activities (e.g., traditional processes), object-centric processes emphasize process data and the various ways users may interact with these data. However, especially in such highly flexible scenarios, the need for analyzing and enhancing corresponding processes persists, and adequate approaches are very much needed.
Object-centric process management has evolved from a theoretical concept to the implementation of the PHILharmonicFlows framework capable of executing object-centric processes. The latter has already been applied in multiple real-world scenarios. This thesis extends the previous research with techniques and algorithms for the innovative analysis and optimization of object behavior as specified in so-called object lifecycle processes. The analysis includes checking the conformance of object lifecycle process models with the actual object behavior recorded in the event log, while at the same time considering flexibility and granularity issues. Additionally, two ways of automatically enhancing object lifecycle processes (i.e., model evolution and instance customization) based on event log information are presented. Moreover, approaches enabling process analysis by deriving event logs from software systems that typically do not provide suitable event data (e.g., legacy software systems), and the discovery of object lifecycle processes from event logs are presented in this thesis. Overall, the approaches presented in this thesis enable a holistic analysis and enhancement of object behavior in the context of object-centric process management.