Discovery of Decision Synchronization Patterns from Event Logs

Tijmen Kuijpers; Karolin Winter; Remco Dijkman

cs.LGMar 20, 2026

Discovery of Decision Synchronization Patterns from Event Logs

Authors: Tijmen Kuijpers, Karolin Winter, Remco Dijkman

Organizations: Eindhoven University of Technology, Department of Industrial Engineering and Innovation Sciences, Eindhoven, The Netherlands

Abstract

Synchronizing decisions between running cases in business processes facilitates fair and efficient use of resources, helps prioritize the most valuable cases, and prevents unnecessary waiting. Consequently, decision synchronization patterns are regularly built into processes, in the form of mechanisms that temporarily delay one case to favor another. These decision mechanisms therefore consider properties of multiple cases at once, rather than just the properties of a single case; an aspect that is rarely addressed by current process discovery techniques. To address this gap, this paper proposes an approach for discovering decision synchronization patterns inspired by supply chain processes. These decision synchronization patterns take the form of specific process constructs combined with a constraint that determines which particular case to execute. We describe, formalize and demonstrate how the constraint for four such patterns can be discovered. We evaluate our approach in two artificial scenarios. First, with four separate process models each containing a single decision synchronization pattern, i.e., we demonstrate that our approach can discover every type of pattern when only this one type is present. Second, we consider a process model containing all four decision synchronization patterns to show generalizability of the approach to more complex problems. For both scenarios, we could reliably retrieve the expected patterns.

Figures & tables

Time	Arrival Value	Max Queue Value	$\frac{Arrival}{Max(Queue)}$	$Max(Queue)$ Enabled	Constraint
5	855	118	7.25	False	False
15	801	855	0.17	True	True
22	222	801	0.28	True	True

Table 2: Observation-Outcome Log for Running Example

Figure 2: Overview Decision Synchronization Pattern Discovery Approach

Pattern	Priority	Blocking	Hold-Batch	Choice
Objective	Prioritization of jobs based on attributes like a high value, faster processing, or a low failure rate	Downstream locations in a process are full or incur high unit holding cost	Reduce unit processing costs of a batch by waiting short time period for more cases to arrive	Avoid depletion of a common resource when two activities require this same resource
Features $\mathcal{F}_{p}$	AttrVal, AttrEnabled	NrTokens	NrTokensEnabled, TimeUntilNext	TimeUntilNext
Constraint $\forall\,i,j\in P,$ $\ i\neq j$	$k_{x}(\text{AttrVal}_{i})\leq a\times$ $k_{y}(\text{AttrVal}_{j})\land$ $k_{x}(\text{AttrEnabled}_{i})$ $==True$ $k_{x},k_{y}\in$ $\{argmin,argmax\}$	$\text{NrTokens}_{i}\leq a$	$\text{NrTokensEnabled}_{i}\geq a$ $\land$ $\text{TimeUntilNext}_{i}\geq b$	$\text{TimeUntilNext}_{i}\geq b$
Process construct $\mathcal{Z}_{p}$ $\exists t_{g},t^{\prime}\in T,\,$ $\exists p_{i},p_{j}\in P$	$(p_{i},t_{g})\in F\land$ $(p_{j},t^{\prime})\in F\land$ $(t^{\prime},p_{i})\in F\land$ $p_{i}\neq p_{j},t_{g}\neq t^{\prime}$	$(t_{g},p_{i})\in F$	$(p_{i},t_{g})\in F$	$(p_{i},t^{\prime})\in F\land$ $(p_{j},t^{\prime})\in F\land$ $(p_{j},t_{g})\in F\land$ $p_{i}\neq p_{j},t_{g}\neq t^{\prime}$

Table 3: Description and Formalization of Decision Synchronization Patterns

Panel of Figure 3 — Figure 3: Process Models per Decision Synchronization Pattern

Figure 4: Decision Tree Priority Pattern

Pattern	Event	Constraint Model	Constraint Discovered
Priority	Job handling	$AttrVal_{arrival}\leq 1.5\times max(AttrVal_{q1})$ $\land$ $argmax(ValEnabled_{q1})=True$	$\frac{AttrVal_{arrival}}{\max(Val_{q1})}\leq 1.497$ $\land$ $argmax(ValEnabled_{q1})=True$
Blocking	Pre-processing	$NrTokens_{q1}<5$	$NrTokens_{q1}\leq 4.5$
Hold- batch	Transportation	$NrTokensEnabled_{q1}>3$ $\land$ $TimeUntilNext_{q1}>2$	$NrTokensEnabled_{q1}>3.5$ $\land$ $TimeUntilNext_{q1}>2$
Choice	Production A	$TimeUntilNext_{arrivalB}>2$	$TimeUntilNext_{arrivalB}>1.915$

Table 4: Single-Pattern Evaluation: Modeled and Discovered Constraints

Figure 5: Fictional Supply Chain Process Model

Pattern	Event	Constraint Model	Constraint Discovered
Priority	Production phone	$AttrVal_{ordered\_pc}\leq$ $max(AttrVal_{stock\_pc})$ $\land$ $argmax(ValEnabled_{stock\_pc})$ $=True$	$\frac{max(AttrVal_{ordered\_pc})}{\max(AttrVal_{stock\_pc})}\leq 50.5$ $\land$ $argmax(ValEnabled_{stock\_pc})$ $=True$
Blocking	Game case arriving	$NrTokens_{stock\_gc}<3$	$NrTokens_{stock\_gc}\leq 2.5$
Hold- batch	Transpor- tation	$NrTokensEnabled_{phone\_NL}>2$ $\land$ $TimeUntilNext_{phone\_NL}>1$	$NrTokensEnabled_{phone\_NL}>2.5$ $\land$ $TimeUntilNext_{phone\_NL}>1.08$
Choice	Production game	$TimeUntilNext_{stock\_pc}>0.5$	$TimeUntilNext_{stock\_pc}>0.5$

Table 5: Multi-pattern evaluation: Modeled and Discovered Constraints

Business Process Modeling

See all 16 papers

Decision-Aware Suffix Prediction and Reasoning of Business Processes
Sep 5, 2026Henryk Mustroph, Stefanie Rinderle-MaPredictive Process Monitoring Process Mining
Technical University of Munich, TUM School of Computation, Information and Technology, Garching, Germany
Resolution limits for process comparison from event data
Sep 17, 2026Antony R. Lee, Peter Tiňo, Iain B. StylesProcess Mining
School of Computer Science, University of Birmingham, Birmingham, B15 2TT, United Kingdom · School of Electronics, Electrical Engineering and Computer Science, Queen’s University Belfast, Belfast, BT7 1NN, United Kingdom
Mind the Long Tail: Understanding the Difficulty of Delay Detection in Business Processes
Aug 14, 2026Keyvan Amiri Elyasi, Lukas Kirchdorfer, Heiner StuckenschmidtPredictive Process Monitoring Process Mining
Data and Web Science Group, University of Mannheim, Germany · SAP Signavio, Walldorf, Germany

Sep 5, 2026J/K move · Enter open · S save

Decision-Aware Suffix Prediction and Reasoning of Business Processes

Henryk Mustroph, Stefanie Rinderle-Ma

Technical University of Munich, TUM School of Computation, Information and Technology, Garching, Germany

Predictive Process Monitoring Process Mining Neuro-Symbolic Reasoning

Open paper AskPDF arXiv

Suffix prediction forecasts the remaining sequence of events of a running case until completion. Most approaches rely on neural networks trained on event logs, which, on average, perform well but struggle with short prefixes or targets belonging to a rare process variant. In such scenarios, the correct path may cross multiple branching decisions, determined primarily by case- and event-level attributes, a signal that NN-based suffix prediction models tend to underweight because they may heavily weight (dense) event labels. Decision mining extracts rules for such decisions from the event log, but has so far been applied only to post-hoc and what-if analysis, not suffix prediction. We therefore extend suffix prediction with decision mining, introducing a decision-aware suffix prediction framework, a neuro-symbolic approach that enables reasoning about predicted events via mined decision rules. Experiments on three of four event logs and three suffix predictors show that the framework can improve suffix prediction, especially for short prefixes but also for rare process variants, and adds intrinsic interpretability.

Abstract

Business Process Modeling

Explore similar work

Decision-Aware Suffix Prediction and Reasoning of Business Processes

Resolution limits for process comparison from event data

Mind the Long Tail: Understanding the Difficulty of Delay Detection in Business Processes