Uncertainty-aware consistency checking in industrial settings
Robbert Jongeling, Antonio Vallecillo · 2023
In this work, we explore how we can assist engineers in managing, in a lightweight way, both consistency and design uncertainty during the creation and maintenance of models and other development artifacts. We propose annotating degrees of doubt to indicate design uncertainties on elements of development artifacts. To combine multiple opinions, we use the fusion operators of subjective logic. We show how these annotations can be used to identify, prioritize, and resolve uncertainty and inconsistency. To do so, we identify the types of design uncertainty and inconsistency to be addressed in two concrete industrial settings and show a prototype implementation of our approach to calculating the uncertainty and inconsistency in these cases. We show how making design uncertainty explicit could be used to tolerate inconsistencies with high uncertainty, prioritize inconsistencies with low associated uncertainty, and uncover previously hidden potential inconsistencies.