Towards a better assessment of event logs quality

Abstract

It is widely observed that the poor event logs quality poses a significant challenge to the process mining project both in terms of choice of process mining algorithms and in terms of the quality of the discovered process model. Therefore, it is important to control the quality of event logs prior to conducting a process mining analysis. In this paper, we propose a qualitative model which aims to assess the quality of event logs before applying process mining algorithms. Our ultimate goal is to give process mining practitioners an overview of the quality of event logs which can help to indicate whether the event log quality is good enough to proceed to process mining and in this case, to suggest both the needed preprocessing steps and the process mining algorithm that is most tailored under such a circumstance. The qualitative model has been evaluated using both artificial and real-life case studies.