A multi-dimensional quality assessment of state-of-the-art process discovery algorithms using real-life event logs

Data-Driven Process Discovery - Revealing Conditional Infrequent Behavior from Event Logs

“…One important challenge for process discovery methods is to handle event logs with noise [2,3]. In practice, event logs often contain noise, e.g., out-of-order events, exceptional behavior, or recording errors [4].…”

“…These techniques are of limited use in real-life settings. Most of the more recent and more sophisticated process discovery methods support noise filtering [3]. Existing noise-filtering methods are based on frequencies [7,8,9,10], machine-learning techniques [11,12], genetic algorithms [13], or probabilistic models [14,15].…”

An efficient process mining model for Petri Nets in process discovery

“…The language based mining technique uses regular expressions to pre structure the input log traces into smaller blocks [11]. The classical synthesis technique based on regions cannot be applied directly because the event log contains only example behavior [12].…”

Discovering Block-Structured Process Models from Incomplete Event Logs

“…Traditionally, models are evaluated with respect to the event log: fitness measures what part of the event log is described by the model, precision is high when the model does not allow too much behaviour that was not in the event log, and generalisation is high when the model allows more behaviour than just the behaviour in the event log. Although fitness, precision, and generalisation are intuitively clear, different formal definitions are possible [13,22,23]. Measuring the quality of a discovered model with respect to its event log might be useful, but whether the best model for the event log is the best model for the system is not captured by these measures.…”