Automated Knowledge Graph Learning in Industrial Processes
L.P. Ammann, Jorge Martínez-Gil, Michael Mayr, Georgios C. Chasparis · Procedia Computer Science · 2025
Industrial processes generate vast amounts of time series data, yet extracting meaningful relationships and insights remains challenging. Knowledge graphs allow to uniquely store and present information, enabling novel capabilities for identifying clusters, temporal relationships, and hidden connections. Building a knowledge graph either requires considerable manual effort and deep expert knowledge of the process or computationally intensive machine learning that provides little transparency in how process structures are identified. This paper introduces a framework for automated knowledge graph learning from time series data, specifically tailored for industrial applications. Moving beyond traditional black-box models process knowledge retrieved from various transparent analytical approaches are amalgamated into a knowledge graph. Similarity analysis is employed to explore simultaneous process relationships, while Granger-causality analysis provides insights into temporally disjoint parameter interactions and causal dependencies between parameters. To illustrate the practical utility of our approach, we present a motivating use case demonstrating the benefits of our framework in a real-world industrial scenario. The user is provided with an intuitive visualization of the process and information that can be enriched with expert knowledge improving decision-making, process optimization, and knowledge discovery.