Business processes are subject to changes due to the dynamic environments in which they are executed. These process changes can lead to concept drifts, which are situations when the characteristics of a business process have undergone significant changes, resulting in event logs that contain data on different versions of a process. The accuracy and usefulness of process mining results derived from such event logs may be compromised because they rely on historical data that no longer reflects the current process behavior, or because the results do not distinguish between different process versions. Therefore, concept drift detection in process mining aims to identify drifts recorded in an event log by detecting when they occurred, localizing process modifications, and characterizing how they manifest over time. This paper focuses on the latter task, i.e., drift characterization, which seeks to understand whether changes unfolded suddenly or gradually and if they form complex patterns like incremental or recurring drifts. However, current solutions for automatically detecting concept drifts from event logs lack comprehensive characterization capabilities. Instead, they mainly focus on drift detection and characterization of isolated process changes. This leads to an incomplete understanding of more complex concept drifts, like incremental and recurring drifts, when several process changes are inter-connected. This paper overcomes such limitations by introducing an improved taxonomy for characterizing concept drifts and a three-step framework that provides an automatic characterization of concept drifts from event logs. We evaluated our framework through elaborate evaluation experiments conducted using a large collection of synthetic event logs. The results highlight the effectiveness and accuracy of our proposed framework and show that it outperforms state-of-the-art techniques.

Steady-state detection (SSD) is a crucial task in the analysis of complex and dynamic systems, as it enables the reliable assessment of system behavior by distinguishing between stable and unstable states. SSD techniques have been extensively studied and applied in various domains, including signal processing and industrial systems. However, their application within the information systems domain, particularly in process mining, has received little attention, even though business processes themselves can be regarded as complex socio-technical systems. In particular, event logs that capture the execution of business processes often contain data from both steady and non-steady states. Mixing up these states can significantly affect the accuracy and reliability of insights from common process mining tasks, such as process performance analysis and process discovery. To address this problem, we propose a dedicated SSD framework for process mining and demonstrate how differentiating between distinct process states can enhance the accuracy and reliability of process mining insights. The SSD framework takes an event log as input and identifies the existing steady and non-steady states along with their corresponding time periods. We evaluate the framework through two experiments: one assessing accuracy using simulated event logs and another demonstrating its impact on three key process mining tasks: process performance analysis, process discovery, and remaining time prediction.