A Deep Dive into the System Knowledge Graphs
by Dirk Fahland, Technical University of Eindhoven (Tu/E)
Preparing data for automated learning of digital twins in circular processes through process mining is challenging:
Source system data is not in right structure, incomplete, and contains errors such as wrongly recorded time-stamps.
Circular processes have the same object re-enter the process it previously finished. For example a medical kit undergoing sterilization after usage in a hospital. your data has to keep track of each execution of the sterilization process and of the kit over its repeated use and sterilization.
Digital twins aimed at optimization need to keep track of process-level control-policies you can optimize. The data has to be ready for this - for example to track how all the medical kits moving from station to station maintain or change their priorities - overtaking others for faster process completion.
This video introduces the System Knowledge Graph and walks you through the fully automated steps of
data integration
enrichment with domain knowledge using inference rules
identification and resolution of data quality problems
modeling distinct process executions and cases in circular processes
modeling system-level behavior of kits for tracking prioritization
and the Auto-Twin Orchestrator that automates the process of learning and maintaining digital twins.
Our technology to build System Knowledge Graphs is available as open-source project https://github.com/AutotwinEU/skg_manager
It is powered by the open-source library PromG which enables process mining using knowledge graphs https://github.com/promg-dev
