8:45 – 10:15
Presenter: Eric Verbeek, Eindhoven University of Technology, The Netherlands & Josep Carmona, Polytechnic University of Catalonia, Spain
Many state-of-the-art automated process discovery techniques, especially the basic ones, struggle to systematically discover accurate and simple process models. In general, as the complexity of the input data increases, the quality of the discovered process models can worsen quickly. Given that oftentimes real-life event logs record very complex and unstructured process behavior containing many repetitions, infrequent traces, and incomplete data, some state-of-the-art techniques turn unreliable and not purposeful. In this presentation, we introduce four advanced process discovery techniques that can be used to overcome the limitations of basic process discovery techniques in the aforementioned scenarios.
10:45 – 12:15
Presenter: Boudewijn van Dongen, Eindhoven University of Technology, The Netherlands
In this advanced session, the students take a closer look into alignments. Using some Petri net theory, we see how alignments can be computed on real models and how the A* algorithm works that is implemented in ProM and many other tools. We then move on to a brief discussion on other conformance metrics beyond fitness, such as precision and generalization, but more importantly, we discuss conformance between a log and a model in the context of a process. We conclude the session with a real example showing true conformance issues with implications in real life.
14:00 – 15:30
Presenter: Jochen De Weerdt, KU Leuven, Belgium
In this session, we will take a deep dive into the topic of preprocessing of event data. While strongly underexposed in research, data preparation in any process mining project will typically require up to 80% of the efforts. As such, significant opportunities exist to research and develop better methods and techniques. We will discuss (1) structure and extraction of event data, (2) event correlation and abstraction, (3) data quality, and (4) more exotic preprocessing techniques such as trace clustering and autoencoders.
swapped from Wednesday to Tuesday
16:00 – 17:30
Presenter: Dirk Fahland, Eindhoven University of Technology, The Netherlands
This session focusses on understanding processes over multiple entities, such as objects, actors and queues. We’ll get to understand the limitations of sequential event logs. Techniques to construct event knowledge graphs from event data over multiple entities will be presented. Finally those will be used to apply process mining over multiple entities for filtering, root-cause analysis, process discovery ad conformance checking.