Conformance checking techniques can be utilized to analyze process executions for their conformance with a process model. Process mining techniques are centered around the notion of a process model that describes the correct behavior of a process. Although, the knowledge about the occurrence of an anomaly is valuable, much more value lies in the knowledge of what was supposed to happen and how to avoid this behavior in the future. While these approaches can accurately pinpoint an anomaly in a process, they do not provide information about what should have been done instead. This technique infers the process solely based on distributions of the execution data, without relying on an abstract definition of the process itself. Process anomaly detection can be used to automatically detect deviations in process execution data. DeepAlign produces better corrections than the rest of the field reaching an overall \(F_1\) score of 0.9572 across all datasets, whereas the best comparable state-of-the-art method reaches 0.6411. We evaluate the performance of our approach on an elaborate data corpus of 252 realistic synthetic event logs and compare it to three state-of-the-art conformance checking methods. DeepAlign utilizes the case-level and event-level attributes to closely model the decisions within a process. By combining the predictive capabilities of both neural networks, we show that it is possible to calculate sequence alignments, which are used to detect and correct anomalies. One is reading sequences of process executions from left to right, while the other is reading the sequences from right to left. At the core of the DeepAlign algorithm are two recurrent neural networks trained to predict the next event. In this paper, we propose DeepAlign, a novel approach to multi-perspective process anomaly correction, based on recurrent neural networks and bidirectional beam search.
0 Comments
Leave a Reply. |