Conference Proceedings

Abstract and Compare: A Framework for Defining Precision Measures for Automated Process Discovery

A Augusto, A Armas Cervantes, R Conforti, M Dumas, M La Rosa, D Reissner

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) | SpringerLink | Published : 2018


Automated process discovery techniques allow us to extract business process models from event logs. The quality of process models discovered by these techniques can be assessed with respect to various quality criteria related to simplicity and accuracy. One of these criteria, namely precision, captures the extent to which the behavior allowed by a discovered process model is observed in the log. While numerous measures of precision have been proposed in the literature, a recent study has shown that none of them fulfils a set of five axioms that capture intuitive properties behind the concept of precision. In addition, several existing precision measures suffer from scalability issues when ap..

View full abstract