A Systematic Literature Review and Classification of Tasks and Datasets in Process Mining

Abstract

Process mining has evolved rapidly over the last 20 years. Starting with process discovery, through conformance checking, to the many different techniques established today. These recent developments also lead to the problem of identifying concrete tasks and their characteristics. A typical difficulty is that tasks with the same characteristics have different names. Furthermore, the evaluation of newly developed algorithms is an important step. In many cases this requires an event log, but not every available event log can and should be used for every task. Two questions arise in this context: First, which process mining tasks exist today? Second, which datasets can be used to evaluate a specific task? The core of this thesis is to provide an answer to these questions. To identify the different process mining tasks, a categorisation is given that describes the different characteristics of the task. In order to answer the second question, a categorisation of the datasets is also made, focusing on which dataset has been used in connection with which task.