Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
While predictive process monitoring commonly treats execution traces in isolation, this work formulates next activity prediction over prefix-Instance Graphs that explicitly incorporate concurrent, contextual process instances. Addressing the limitation of standard sequence baselines that ignore environmental and cross-instance dynamics, the authors evaluate multiple graph encoding strategies fed into Graph Neural Networks. Across multiple real-world event logs, the results show that explicitly modeling structured contextual dependencies provides consistent performance gains over context-agnostic representations.
Workflow events do not execute in a vacuum: capturing concurrent process context via relational instance graphs consistently outperforms standard, isolated trace modeling in next-step predictions.
Predictive process monitoring aims at forecasting various aspects of running processes. Among the different tasks, next activity prediction represents the most extensively investigated. However, only a limited number of existing approaches explicitly encode contextual information, i.e., the environmental conditions in which the process is executed, typically modeled through event log attributes or aggregated measures. In this paper, an approach based on the concept of Instance Graphs is introduced. To incorporate contextual process instances, several encoding strategies are proposed and evaluated by measuring their impact on prediction performance. For each encoding strategy, a set of prefix-Instance Graphs is generated and subsequently provided as input to a Graph Neural Network for the classification task. The proposed approach is evaluated on multiple real-world event logs, and the experimental results demonstrate that incorporating contextual process instances benefits prediction performance.