abstract
- In Internet of Things (IoT)-based process monitoring, the abstraction of sensor data into process events is often ambiguous. This ambiguity results in multiple plausible interpretations of the resulting event log, hindering process monitoring. We propose a novel algorithm that leverages alignment-based conformance checking to disambiguate process events in an online monitoring setting. For each ambiguous event, the algorithm explores the state space up to a k-bounded depth to identify the interpretation with the local minimum number of mismatches as the most plausible event interpretation. To make the approach feasible at run time, we employ an FM-index-based technique that enables fast computation of local all optimal alignments. A prototypical implementation demonstrates suitability for online disambiguation in IoT scenarios.