Object-Centric Process Mining


Organizational Information

  • Coordinator: Niklas Adams
  • The registration is carried out via the central registration process on the web site SUPRA maintained by the Computer Science Department.
  • This seminar is held in English (all written reports and presentations will be in English).

Only the students who will participate in the seminar will be contacted and they will be provided with the kick-off meeting date and time. All deadlines for submission of the outline, term paper, and presentation slides will be announced in the kick-off meeting. Presentations will take place on probably two or three days and this will also be announced in the kick-off meeting.


Process mining offers algorithms to analyze the event data generated during the execution of processes. Traditional process mining assumes an event log where every event has an activity, a timestamp, and one case identifier that identifies the end-to-end run through the process. However, this assumption is often not met in reality: Events can be related to multiple case identifiers, e.g., in a production process where different components are produced and later assembled. If events are related to multiple case identifiers, so-called objects, the event data needs to be transformed to meet the assumptions of traditional process mining, i.e., events with only one case identifier. This process is called flattening and it introduces data quality issues to the flattened event log that will influence the accuracy of the process analysis.

Object-centric process mining is a field of process mining that addresses this problem. Object-centric process mining algorithms are applied to the event data with multiple case notions, i.e., object-centric event data, to analyze a process without the negative influence of flattening. Since the impact of flattening can be quite detrimental to several process mining tasks, the topic of object-centric process mining has gained traction over the last years, with Celonis introducing an object-centric process model, Celosphere, academics establishing file standards for exchange and storage of object-centric event logs, and practitioners reporting increased accuracy of process analysis using object-centric process mining. This seminar intends to address the latest advances in object-centric process mining.


In this seminar, students will learn two major skillsets for scientific working: Scientific communication and scientific content. We will start with learning how to structure and communicate scientific work to highlight the researcher's contributions. This will be applied to the domain of object-centric process mining, equipping the students with state-of-the-art scientific knowledge. The students will showcase the learned skills by writing a short research paper on an assigned topic. This includes a literature review, formalization, and a (replication, adaption, or –if infeasible– translation of an) evaluation. The students will give a presentation about their research paper.


Basic knowledge of Computer Science and a solid mathematical background are needed (e.g. set theory, logic, automata, etc.). Knowledge in process mining, i.e., process discovery, and conformance checking, is not necessarily required but highly encouraged. Please state your current knowledge when you apply for the seminar.


The registration is carried out via SUPRA. Please state your qualifications and prior knowledge when applying.