How does schema.org usage fit into Linked Data principles?

Question

I am starting to learn schema.org schema. I come from the RDF/OWL community.

To my surprise I could not indeed find an RDFS or OWL vocabulary representation of schema.org. More specifically, I realised that it was just a HTML page describing the schema. I also found an RDFa representation of the schema. Not sure that it can be a legal RDFS representation by the way. All of this got me a bit confused.

In the meantime I found the following works: http://topbraid.org/schema/ and http://schema.rdfs.org/.

Hence I have the following few questions:

• 1- Are both work achieving the same thing? if not what is the difference between the two?

• 1.2- What is their respective goal, and use case?

• 3- The URIs do not correspond to the URI of the ontology, isn’t it odd? How does that align itself with Linked Data principles?

• 4- Is schema.org fully aligned with Linked Data principle?

• 5- I find it odd to reference a schema that has no RDFs representation? How one can one use it in tools like Protégé for instance?

I would really appreciate to have some clarification over this.

Hakeem

duplicated at answers.semanticweb.com/questions/28127/… – Sennar

Best practice (for publishing RDF Ontologies) suggests, that an HTTP(S) GET request would be answered with the ontology in the RDF serialization format requested in the HTTP content-type header, on the actual schema URL (in this case: "http://schema.org/"). Sadly, most Onotologies don't follow this practice, including schema.org. – Cerelia

Answer 1

From http://schema.org/docs/datamodel.html:

The canonical machine representation of schema.org is in RDFa:
schema_org_rdfa.html

(See my answer on Webmasters SE for other representation.)

This HTML+RDFa version uses the RDFS vocabulary. It’s as good as any other RDF serialization. If you don’t like RDFa, you could of course use your favorite converter to get Turtle, RDF/XML, or whatever you prefer.

Suh

Thanks to both of your answer. I understand better. I guess the only thing that is not obvious when you come from extensive usage of ontology trough protege, etc... is: where is the vocabulary that i can use to produce my data file ? I was looking to import it in protege but could not do it. so now i get it, if i want to do it, i must extract the RDFA first and import it in protege. – Hakeem

One last question, this canonical version of RDFa is rdfs. Hence i was wondering, if importing it in a owl2-dl ontology make it owl-full: the typical case encountered with the dublin core ontology for instance.In other words does it use only rdfs construct that are compatible with owl2-dl ? – Hakeem

@Hakeem Well, it won't be a legal OWL 2 DL ontology, at any rate, since it will be missing the declaration axioms that declare each property to be a object property, datatype property, or annotation property. – Sennar

Answer 2

When schema.org was launched in June 2011, it didn't include any up to date schema as RDFS or OWL. http://schema.rdfs.org/ was created to bridge that gap. Since then, the schema.org folks created http://schema.org/docs/schema_org_rdfa.html which is the canonical schema used to generate all the HTML pages for the schema.org types and properties (so it's always up to date). You should use that schema if you intend to publish Linked Data using the general schema.org schema.

1. Not sure what you mean. The references to types and properties from are all linking to the schema.org type and property URIs, for example http://schema.org/CreativeWork.

2. Linked Data is mostly about linking data items to others data items from the same dataset and if possible to external datasets too (see Linked Open Data project).

3. RDFa is one way to serialize RDF data. Therefore http://schema.org/docs/schema_org_rdfa.html is a fine representation of schema.org in RDF. You can the RDF/XML or Turtle if you like from this RDFa schema, by using your favorite parser.

Moonlit