Events play an important role in information systems engineering to facilitate the analysis of relevant happenings in a system or business process via process mining. The Internet of Things (IoT) provides new ways of collecting execution-related events in real world–physical–systems using sensors and actuators. However, these new sources emit data at a too fine-grained level, which prevents process mining from deriving meaningful insights. We present a generic event abstraction framework to lift low-level data to higher level events. Starting with annotated IoT data, we generate stream processing applications that encode change patterns derived from the low-level data. These applications are then used for detecting events and activities at runtime within our proposed software architecture. We evaluate the approach for process executions in smart manufacturing and smart healthcare.
Seiger et al. (Thu,) studied this question.