Enhancement in Process Mining Model by Repairing Noisy Behavior in Event Log

Abstract

Abstract Companies and organization aim to be competitive in order to improve the performance of the business process. In recent years, process mining attracts researchers and their main purpose is to extract accurate information from the process-related data. Enhancement is one of their main types, which is used to prolong or enhance/improve the existing process by taking the information from the actual process of an event log. As we know, enhancement has two types Extension and Repair. In our paper, we use the repair type of enhancement. It is common practice that logging errors in information systems or the presence of special behavior mean that we have the actual event log with the noise. In our paper, we investigate the process mining model with the presence of noise in the event log. We repair the event logs by decomposing them into many sub-logs and removing the noise behavior in the sub-logs by using covering probability. Repaired sub-logs are than added to the original event log at the appropriate place. We also propose in our research work a probabilistic method that depends on frequency occurrence for activities in a particular situation. The proposed method allows us to remove noisy and abnormal behavior in the event log, which permit us to gain a over-all perspective in the process Hence we generate artificial event logs with the presence of noise behavior in the ProM framework. By using RapidMiner based ProM Extension, we generate a test set that demonstrates, that we can find out and repair multiple types of noisy behavior in an event log. Therefore the proposed method also clearly shows that after repairing the event log we can improve the performance of a process mining model.