Trace۲Vec-CDD: A Framework for Concept Drift Detection in Business Process Logs using Trace Embedding

Business processes are subject to changes during their execution over time due to new legislation, seasonal effects, and so on. Detection of process changes is alternatively called business process drift detection. Currently, existing methods unfavorably subject the accuracy of drift detection to the effect of window size. Furthermore, most methods have to struggle with the problem of how to select appropriate features specifying the relations between traces or events. This paper draws on the notion of trace embedding to propose a new framework (Trace۲Vec CDD) for automatic detection of suddenly occurring process drifts. The main contributions of the proposed approach are: (i) It is independent of windows. (ii) Trace embedding, which is used for drift detection, makes it possible to automatically extract all features from relations between traces. (iii) As attested by synthetic event logs, our approach is superior to current methods in respect of accuracy and drift detection delay.

Keywords:

Process mining , Concept drift , Process changes , Word embedding