Online Approximate Conformance Checking

Online conformance checking is used to verify the compliance between the event stream and the given model. The existing methods need to detect each event in the event stream, which not only increases requirements for reference models but also low operating efficiency. Therefore, this paper proposes an approximate conformance checking method that does not need to build a model. We need to cluster the existing logs, select representative traces to build a model support set, and calculate the upper and lower bounds of fitness as the basis of conformance judgment. During the online operation of the system, the event stream is temporarily stored in the event stack, and the event sequence is only taken out at the key activity nodes for conformance checking with the model support set. Finally, the performance of the proposed method is evaluated through a practical log. After comparing with the existing methods, the results show that our algorithm is feasible and has high accuracy.