Me necessary to complete an activity and contextual facts, for instance employee facts and client transactions extracted from the ERP method of a medium-sized logistics business. The distances involving situations have been calculated using the variations amongst adjusted durations. Tax et al. [31] proposed a framework for the automated generation of label refinements primarily based on the time attribute of events, permitting to distinguish behaviorally distinctive situations in the similar event kind primarily based on their time attributes. The events generated by 1 sensor have been clustered employing a mixture model consisting of elements of theAppl. Sci. 2021, 11,10 ofvon Mises distribution, that is the circular equivalent with the typical distribution. Four tactics had been applied for various label refinements on three event logs from the human behavior domain. Song et al. [28] proposed an approach primarily based around the minimum alter principle to repair timestamps that do not conform to temporal constraints, e.g., to locate a repair that is definitely as close as you possibly can to the original observation. The problem is tackled by identifying a concise set of promising candidates applying an algorithm for computing the optimal repair from the generated candidates, as well as a heuristic approximation by picking repairs from the candidates. Rogge-Solti et al. [32] presented a process to repair the timed event logs by combining stochastic Petri nets, alignments, and Icosabutate Epigenetics Bayesian networks. The method decomposes the problem into two sub-problems: (a) repairing the time and (b) repairing the structure for each trace. This perform requires all the observed data into account and gets effective estimations for the activity durations and path probabilities. Fischer et al. [33] proposed an strategy for detecting and quantifying timestamp imperfections in event logs based on 15 quality metrics structured along 4 data excellent dimensions and log levels. three.2.two. Detection isualization Procedures Detection isualization strategies aim to recognize, group, and isolate those events or traces that may generate complications in the good quality on the event log. Within this group, two approaches are identified: clustering and pattern-based tactics. Clustering techniques divide the event log into several subsets, facilitating the understanding and evaluation of every single member in the subsets. Then, the subsequent step is the identification of noise/anomalous elements within the analyzed subsets. Clustering is one of the strategies most used for data preprocessing in method mining, which has been mainly utilised for the identification of high-quality problems IQP-0528 Biological Activity related with noisy values, also as information diversity. From the formation of related clusters, it truly is attainable to identify imperfection patterns related to noisy information inside the diverse attributes from the event logs. Several techniques have been proposed inside the last decade for trace clustering. They are able to be divided into 3 approaches: vector space approaches [347], context aware approaches [382], and model-based approaches [438]. Most of the clustering algorithms aforementioned think about only the event log as input, and use various internal representations for producing the clusters. Traditionally, these algorithms happen to be applied without taking into consideration the availability of a approach model. In contrast, in current functions [49,50], a different view on traces clustering of an event log is presented. The authors assume that a procedure model exists and it is employed to construct simpler groupings of.