RDF Schemas Have Errors

[joshhansen]

I'm referring to the schemas located at

• http://www.gedcomx.org/model/gx.rdf.xml
• http://www.gedcomx.org/model/gxr.rdf.xml
• http://www.gedcomx.org/model/gxc.rdf.xml

  Problems

  1. The general and record model schemas don't validate. See [1] and [2]. The cause for this is that some rdfs:comment properties contain not just text but embedded HTML tags (<i> and <a>). One remedy is to surround these HTML fragments with <![CDATA[...]]> tags and to declare their datatype to be XMLLiteral. This is demonstrated in this diff.
  2. None of the namespaces nor any of the vocabulary defined therein resolve to RDF as is standard practice, and as specifically mandated in the Linked Data paradigm. The namespaces in use seem to be http://gedcomx.org/, http://gedcomx.org/conclusion/v1/, and http://gedcomx.org/record/v1/. Except for http://gedcomx.org/ itself, they all give 404's as far as I can tell. If http://gedcomx.org/ is to be used as a namespace, it should serve either HTML or RDF as requested using HTTP content negotiation.
  3. In the conclusion model, both the original and formal properties have a domain of formalizeable. However, formalizeable is never defined. Additionally, since the domain of a property is supposed to be a class, this should probably be capitalized as Formalizeable.
  4. Incorrect file extension. .rdf.xml is not the right file extension for RDF/XML documents. The correct extension is .rdf. (Nitpicky, I know....)

     Potential Problems

The following things seem a little iffy to me:

1. In the general model, the persistentId and source properties have no domain. Maybe this is intentional?
2. In the conclusion model, the father, fatherRelationship, mother, motherRelationship, spouse, and spouseRelationship properties have no domain. I doubt this is intented.
3. In the record model, the alternateId and persistentId properties have no domain. Again, maybe on purpose?
4. The schemas do not use XML namespaces for the very namespaces they are defining. For example, instead of http://gedcomx.org/record/v1/Age it would be more concise (and more readable, in my opinion) to say gxr:Age using the namespace prefix of gxr to represent the record model namespace.
5. There are a ton of top-level classes in gx.rdf.xml. Shouldn't they be part of a hierarchy?
6. I don't know OWL that well, but a number of the classes in gx.rdf.xml represent conditions that I suspect could be expressed using OWL. For example, NeverHadChildren. Just a thought.

[stoicflame]

Excellent. These should be pretty easy to fix. Thanks for the report.

[stoicflame]

(Doing some triage work.)

We are no longer using RDF in the existing specifications, and therefore are no longer publishing the RDF schemas. So I'm closing this issue. Maybe someday we can provide an RDF representation of GEDCOM X data, but this work is now outside the scope of this current project.