Process mining includes the automated discovery of processes from event logs. Based on observed events (e.g., activities being executed or messages being exchanged) a process model is constructed. One of the essential problems in process mining is that one cannot assume to have seen all possible behavior. At best, one has seen a representative subset. Therefore, classical synthesis techniques are not suitable as they aim at finding a model that is able to exactly reproduce the log. Existing process mining techniques try to avoid such "overfitting" by generalizing the model to allow for more behavior. This generalization is often driven by the representation language and very crude assumptions about completeness. As a result, parts of the model are "overfitting" (allow only for what has actually been observed) while other parts may be "underfitting" (allow for much more behavior without strong support for it). None of the existing techniques enables the user to control the balance between "overfitting" and "underfitting". To address this, we propose a two-step approach. First, using a configurable approach, a transition system is constructed. Then, using the "theory of regions", the model is synthesized. The approach has been implemented in the context of ProM and overcomes many of the limitations of traditional approaches.
Process discovery algorithms typically aim at discovering process models from event logs that best describe the recorded behavior. Often, the quality of a process discovery algorithm is measured by quantifying to what extent the resulting model can reproduce the behavior in the log, i.e. replay fitness. At the same time, there are other measures that compare a model with recorded behavior in terms of the precision of the model and the extent to which the model generalizes the behavior in the log. Furthermore, many measures exist to express the complexity of a model irrespective of the log.In this paper, we first discuss several quality dimensions related to process discovery. We further show that existing process discovery algorithms typically consider at most two out of the four main quality dimensions: replay fitness, precision, generalization and simplicity. Moreover, existing approaches cannot steer the discovery process based on user-defined weights for the four quality dimensions. This paper presents the ETM algorithm which allows the user to seamlessly steer the discovery process based on preferences with respect to the four quality dimensions. We show that all dimensions are important for process discovery. However, it only makes sense to consider precision, generalization and simplicity if the replay fitness is acceptable.
Abstract. The holy grail in process mining is an algorithm that, given an event log, produces fitting, precise, properly generalizing and simple process models. While there is consensus on the existence of solid metrics for fitness and simplicity, current metrics for precision and generalization have important flaws, which hamper their applicability in a general setting. In this paper, a novel approach to measure precision and generalization is presented, which relies on the notion of anti-alignments. An anti-alignment describes highly deviating model traces with respect to observed behavior. We propose metrics for precision and generalization that resemble the leave-one-out cross-validation techniques, where individual traces of the log are removed and the computed anti-alignment assess the model's capability to describe precisely or generalize the observed behavior.
Abstract. Process mining techniques attempt to extract non-trivial knowledge and interesting insights from event logs. Process models can be seen as the "maps" describing the operational processes of organizations. Unfortunately, traditional process discovery algorithms have problems dealing with less-structured processes. Furthermore, existing discovery algorithms do not consider the analyst's context of analysis. As a result, the current models (i.e., "maps") are difficult to comprehend or even misleading. To address this problem, we propose a two-phase approach based on common execution patterns. First, the user selects relevant and context-dependent patterns. These patterns are used to obtain an event log at a higher abstraction level. Subsequently, the transformed log is used to create a hierarchical process map. The approach has been implemented in the context of ProM. Using a real-life log of a housing agency we demonstrate that we can use this approach to create maps that (i) depict desired traits, (ii) eliminate irrelevant details, (iii) reduce complexity, and (iv) improve comprehensibility.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.