Artificial Datasets For "Online Conformance Checking Using Behavioural Patterns"

Dataset containing artificial datasets for the stress test of the online conformance prototype and for the correlation of the results of the online conformance checker with state of the art technique. <strong>Stress test log</strong> We randomly generated a BPMN model containing 64 activities and 26 gateways. The model was then used to simulate an event stream of 2 million events. <strong>Correlation logs</strong> We generated 12 random process models with number of activities according to a triangular distribution with lower bound 10, mode 20, and upper bound 30. We did not include duplicate labels, a probability of 0.2 for addition of silent activities, moreover, the probability of control-flow operator insertion was: 0.45 for sequence, 0.2 for parallel and xor-split operators, 0.05 for an inclusive-or operator and 0.1 for loop constructs. Incremental noise levels (both on a trace- and event-level) were introduced in the logs. Probability of trace- and event-level noises ranged from 0.1 to 0.5 with steps of 0.1.