How well did it recover? Impact-aware conformance checking
- PDF / 1,203,735 Bytes
- 25 Pages / 439.37 x 666.142 pts Page_size
- 55 Downloads / 205 Views
How well did it recover? Impact-aware conformance checking Arava Tsoury1 · Pnina Soffer1 · Iris Reinhartz-Berger1 Received: 8 December 2019 / Accepted: 16 October 2020 © Springer-Verlag GmbH Austria, part of Springer Nature 2020
Abstract Alignment-based conformance checking techniques detect and quantify discrepancies between process execution and the expected behavior as depicted in process models. However, often when deviations from the expected behavior occur, additional actions are needed to remedy and restore the process state. These would seem as further reducing conformance according to existing measures. This paper presents a conformance checking approach which considers the response to and recovery from unexpected deviations during process execution, by analyzing the data updates involved and their impact on the expected behavior. We evaluated our approach by applying it to a reallife case study, utilizing different baseline alignment techniques. The results show that our approach succeed to capture adapted behavior in response to deviations and may provide insights concerning the implications of deviations in the process. Keywords Conformance checking · Alignment · Data impact analysis · Business processes Mathematics Subject Classification 93-04
1 Introduction Business process models depict the standard and common course by which processes are expected to be executed. Over the years, process mining techniques [31] have been developed, utilizing event logs, which commonly exist in information systems
B
Arava Tsoury [email protected] Pnina Soffer [email protected] Iris Reinhartz-Berger [email protected]
1
University of Haifa, 3498838 Mount Carmel, Haifa, Israel
123
A. Tsoury et al.
and record activities performed using the system. One of the main application areas of process mining is conformance checking [9], which attempts to measure the extent to which the actual behavior, reflected in an event log, conforms to or matches the expected behavior, specified in a process model. Most conformance checking techniques (e.g., [1, 2, 30]) focus on control flow aspects, addressing the order and flow of activities in the process. They measure fitness of a model and an event log, namely, the ability of the model to explain the execution of a process as recorded in the given event log. Additional aspects, besides control flows, have also been considered for conformance checking (e.g., [11, 13, 24, 25]), including resources, time, and data-related rules. Although conformance checking uses the modeled process as a baseline for comparison, process models rarely capture the full range of possible behaviors. In particular, exceptions that may occur and possible compensation that may be needed due to errors are typically not described in the process model [4]. Hence, comparison in such cases may reveal deviations from the prescribed process model, as well as incorrect changes of data. These may have further consequences as the process continues, involving additional activities that may not be specified
Data Loading...
