Knowledge Graph Primer: Creating a digital twin graph model

A series of tutorials designed to introduce the use of knowledge graphs to an industrial environment (process, manufacturing industries; construction etc.) to technical users with some database experience but not necessarily experience of the use of graphs.

Throughout the tutorials, there is an example ‘thread’ of building models of process plants and equipment with the objective of creating a fit-for-purpose^[1] ‘digital twin’ model. This thread culminates with a ‘recipe’ for building fit-for-purpose digital twin models, particularly in the manufacturing and equipment domains.

Each tutorial is illustrated by the de

Session 1: Introduction to graphs

Graph databases, how they fit in the database landscape, their advantages, and disadvantages.

1. What is a knowledge graph database
2. Why is it different,
3. What are its advantages
4. What are its disadvantages  
5. How to build a graph database
   1. Triplestores
   2. Development tools

Session 2: Simple modeling with graphs

The tools required to build a fit-for-purpose model: RDF, RDFS

1. Using RDF, RDFS, and OWL (maybe) to describe graph models
2. Creating 1-D models with graphs
3. Querying graphs
   1. with SPARQL
   2. with OData

Session 3: Creating a realistic plant model: a digital-twin

Tackling the real requirements of a fit-for-purpose digital-twin model

1. What are the issues to be solved applying knowledge graphs to an industrial application
2. Limitations of 1D modeling (simple subject-property-object)
3. Useful concepts from Basic Formal Ontology
4. Extension to 2D modeling (reified statements so we can have other data such as Units-of-measure, accuracy, version and so on)
5. Extension to 3D modeling (so we can capture the fact that knowledge is always changing over time)
6. Avoiding complex taxonomies with shapes

[1] Fit-for-purpose: Although focused on the manufacturing and equipment domains, we want to build a model which the participants can relate and easily transfer to their own problems. This model should be able to answer the detailed questions that they know will be required in practice: change management, moving equipment, lifecycle of equipment, missing values, measurement values, units-of-measure and so on.