SC²S Colloquium - November 2, 2017
|Date:||November 2, 2017|
|Time:||4:00 pm, s.t.|
Sebastian: Creating a Binary labeled Event Log for Process Mining using Semi-Supervised Log Pattern Detection and Exploration
The idea of creating a binary labeled event Log for Process Mining is to include small-scale user input which partially describes wanted and unwanted behavior. The input is used to detect pattern which defines the wanted and unwanted behavior. This patterns are semi-automated enriched by data-aware splits to fully define the event log labels. This extends the existing process discovery techniques with the detection of frequent unwanted behavior. The benefit is that on this labeled event log we can run classical process discovery to extract models for both wanted and unwanted behavior. The two output models then are used for further analyses, the positive model as definition of the desired process in conformance checking and the negative model as the indicator for opportunities to improve the process.