Event Log Extraction for the Purpose of Process Mining: A Systematic Literature Review

Abstract

Process mining bridges the gap between process model analysis and data-oriented analysis, by enabling automated discovery of process models, comparison of existing process models with an event log of the same process and improvement of existing process models. Process mining prerequisite is an information system that supports and controls real-life business processes and consequently stores event data, such as messages, transactions, and logs, as event logs in some type of a database. Event data is then extracted, filtered, and loaded into process mining software, where a certain type of process mining can be conducted. Process-aware information systems (PAIS), which assume an explicit notion of a case to correlate events of a process, provide such logs directly. However, many information systems that support execution of business processes are not explicitly process-aware and due to the variability of the event data sources, this phase of process mining is challenging and the most time-consuming. Consequently, various event log extraction techniques, approaches, and tools are being developed, both specific and generic. To make a contribution to the issue, this paper presents a systematic literature review conducted with the aim to answer the questions about genericity of the approaches, applicability by non-experts, and developed feasible tools.