Edit

Convert industry-standard ontologies to DTDL for Azure Digital Twins

Most ontologies are based on semantic web standards such as OWL, RDF, and RDFS.

To use a model with Azure Digital Twins, you must convert it to DTDL format. This article describes general design guidance in the form of a conversion pattern for converting RDF-based models to DTDL so that you can use them with Azure Digital Twins.

Although the examples in this article are building-focused, you can apply similar processes to standard ontologies across different industries to convert them to DTDL as well.

Conversion pattern

There are several third-party libraries that can be used when converting RDF-based models to DTDL. Some of these libraries allow you to load your model file into a graph. You can loop through the graph looking for specific RDFS and OWL constructs, and convert them to DTDL.

The following table is an example of how RDFS and OWL constructs can be mapped to DTDL.

RDFS/OWL concept RDFS/OWL construct DTDL concept DTDL construct
Classes owl:Class
IRI suffix
rdfs:label
rdfs:comment
Interface @type:Interface
@id
displayName
comment
Subclasses owl:Class
IRI suffix
rdfs:label
rdfs:comment
rdfs:subClassOf
Interface @type:Interface
@id
displayName
comment
extends
Datatype Properties owl:DatatypeProperty
rdfs:label or INode
rdfs:label
rdfs:range
Interface Properties @type:Property
name
displayName
schema
Object Properties owl:ObjectProperty
rdfs:label or INode
rdfs:range
rdfs:comment
rdfs:label
Relationship type:Relationship
name
target (or omitted if no rdfs:range)
comment
displayName

The following C# code snippet shows how an RDF model file is loaded into a graph and converted to DTDL, using the dotNetRDF library.

using VDS.RDF.Ontology;
using VDS.RDF.Parsing;
using System;
using System.Collections.Generic;
using Newtonsoft.Json;

namespace DigitalTwins_Samples
{
    public class DigitalTwinsConvertRDFSample
    {

        public void Run()
        {

            Console.WriteLine("Reading file...");

            FileLoader.Load(_ontologyGraph, rdfFile.FullName);

            // Start looping through for each owl:Class
            foreach (OntologyClass owlClass in _ontologyGraph.OwlClasses)
            {

                // Generate a DTMI for the owl:Class
                string Id = GenerateDTMI(owlClass);

                if (!String.IsNullOrEmpty(Id))
                {

                    Console.WriteLine($"{owlClass.ToString()} -> {Id}");

                    // Create Interface
                    var dtdlInterface = new DtdlInterface
                    {
                        Id = Id,
                        Type = "Interface",
                        DisplayName = GetInterfaceDisplayName(owlClass),
                        Comment = GetInterfaceComment(owlClass),
                        Contents = new List<DtdlContents>(),
                    };

                    // An OWL graph can have parent/child classes.
                    // So to understand if an OWL class is a base class or a child class,
                    // look for a superclass on any given OWL class.
                    // If found, convert these to parent + child Interfaces using DTDL extends.
                    IEnumerable<OntologyClass> foundSuperClasses = owlClass.DirectSuperClasses;

                    //...
                }

                // Add interface to the list of interfaces
                _interfaceList.Add(dtdlInterface);
            }

            // Serialize to JSON
            var json = JsonConvert.SerializeObject(_interfaceList);

        }
    }
}

Next steps

Continue on the path for developing models based on ontologies: Full model development path.