Assessing the Impact of Context Data on Process Outcomes During Runtime

Abstract

The outcome of a process e.g., the quality of a produced part, constitutes a key performance indicator for process analysis and monitoring. Process outcomes are not only affected by process data, but also by data that is not associated with the process logic through decisions or task input. The rising temperature in a machine, for example, might cause deterioration of part quality. Assessing the impact of context data on the process outcome at runtime is particularly useful to reduce the reaction time to possible errors or deviations. However, as process models contain loops and decisions, grouping and making context data streams interpretable is not always straight-forward, especially under the condition that describing dependencies between context data and process data should be simple and flexible. The contribution of this paper is a classification of context data types, how they are connected to a process model, and how process models can be segmented into stages to group semantically related tasks. The impact of context data on the process outcome is then determined during runtime, i.e., as a process instance is progressing through these segments at runtime, impact calculations using context data can be gradually refined. The approach is prototypically implemented and applied to an artificial logistics and a real-world manufacturing data set.

This work has been partly funded by the Austrian Research Promotion Agency (FFG) via the “Austrian Competence Center for Digital Production” (CDP) under the contract number 881843. This work has been supported by the Pilot Factory Industry 4.0, Seestadtstrasse 27, Vienna, Austria.