Process mining techniques bridge the gap between data science and process management by extracting insights into process behavior from event logs. Traditional process mining methodologies typically assume deterministic and noise-free event data, yet real-world scenarios often involve stochastic effects, uncertain information, incomplete traces, and partial observability of events. Consequently, there is a growing need for methods that accommodate stochastics, uncertainty, and non-determinism in process mining.
Recent advancements reflect that many processes cannot be fully described by linear, totally ordered sequences of events. Instead, underlying partial-order constraints or concurrency considerations frequently dictate the actual process flows. Further, data arising from such processes may be incomplete, ambiguous, or inherently uncertain, requiring models that account for various degrees of randomness and imprecision.
This workshop aims to bring together researchers and practitioners who are interested in extending the current scope of process mining to incorporate stochastic modeling (through, e.g., stochastic Petri nets, Markov processes), and the handling of uncertainty and non-determinism in event data. By focusing on these topics, we can explore the frontiers of dynamic, real-world processes, generating robust, real-time insights and enabling more accurate predictions and decision-making.
The goal of the First International Workshop on Stochastics, Uncertainty, and Non-determinism in Process Mining (SUN-PM) is to promote the expansion of existing research by providing a platform to discuss novel techniques, theories, and applications in the realm of process mining when data or models exhibit uncertainties, stochastic characteristics, or partial-order structures. The workshop will host new theoretical contributions in the aforementioned topics, and bring together academic researchers and industry experts to exchange ideas, showcase innovative methods, and identify challenges and future directions.
Accepted full-length papers will be submitted for publication in a volume of the Lecture Notes on Business Information Processing series by Springer.
Depending on the number of submitted and accepted papers, the authors of a selection of best full-length papers may be invited to submit an extended version for publication in Springer's Process Science journal.
Topics of Interest
We invite contributions that explore or leverage stochastic, uncertain, or non-deterministic methods, as well as partial-order semantics in process mining. Submitted contributions need to be original and unpublished papers. Topics of interest include, but are not limited to:
- Process discovery, at which either the input log is implicitly or explicitly uncertain (e.g., noise, logs generated by LLMs of stochastic models), or the output model contains an explicit notion of stochastics or uncertainty
- Conformance checking techniques on stochastic models, uncertain logs, or any combination thereof
- Extensions of quality or distance metrics, and KPIs on stochastic models and/or uncertain event logs
- Process simulation, the influence of stochastic model quality on simulation, discovery of simulation models, quality measures for simulation models, etc.
- Approximate, probabilistic or non-deterministic methods for declarative constraint checking
- Process discovery techniques for partially ordered, probabilistic, or uncertain data
- Conformance checking techniques for partially ordered, probabilistic, or uncertain data
- New types of analyses that have a stochastic, uncertain or non-deterministic flavour
- New ways of modeling stochastic, uncertain, or non-deterministic behavior and related learning approaches
- Other types of non-deterministic modelling, analyses or techniques, such as causal aware process mining, or statistical tests for process behavior
- Applications or case studies in which stochastics, uncertainty or non-determinism play an important role
Workshop Keywords
Process Mining, Stochastic Process Models, Uncertain Event Data, Non-determinism, Partially-Ordered Event Data