abstract
- Internet of Things (IoT) technologies can be leveraged to monitor and check the conformance of business process executions in the absence of a Business Process Management (BPM) System. This poses the challenge of detecting process activities from low-level sensor data by means of event abstraction. Existing methods rely on fully supervised approaches, combining sensor data and process event logs to train classification models. However, frequently these logs are not available, invalidating the applicability of such methods. In this paper, we propose a semi-automated approach to detecting process activities from sensor data based on frequently repeated subsequences and automata. We evaluate the approach with a proof-of-concept implementation and data from a small-scale smart factory. Our evaluation demonstrates the applicability of the proposed approach and its effectiveness in detecting process activities from sensor data to support domain experts with data analysis through automated suggestions. © 2025 Elsevier B.V.. All rights reserved.