Differences between rdf:resource, rdf:about and rdf:ID

Question

What are the conceptual differences between rdf:resource, rdf:about, and rdf:ID. I did some investigation but the difference between them are not clear for me yet. For example, whether rdf:ID is used when declaring a resource for the first time, rdf:resource is used for referencing an already existing resource, etc.

I would be glad if you provide some little examples.

Answer 1

To be clear this is only about a particular way of writing rdf: namely RDF/XML. Other syntaxes don't feature these differences.

With that disclaimer out of the way:

What we're trying to do is write statements of the form:

subject predicate object 

and in particular:

subjectURI predicate objectURI 

So how do we introduce subject and object URIs in RDF/XML?

1. rdf:about sets the subject URI of a statement, which may be absolute (http://example.com/) or resolved relative to the BASE of the document (e.g. /foo/bar, #frag). (Like href in html)
2. rdf:resource sets the object URI of a statement, once again either absolute or relative.
3. rdf:ID sets the subject URI, but it can only be within this document. An ID can also only be used once. Very like <a name="baz"> or id="baz" in html.

rdf:ID is discouraged since

1. you can replace it with an rdf:about or rdf:resource with a fragment #baz and
2. it can cause xml issues if you use the same id more than once.

That is, it's redundant and a potential source of errors.

In retrospect there typically only needs to be one attribute to specify a URI, since whether something is a subject or object is apparent from the RDF/XML syntax:

<ex:Foo ...> - subject   <ex:prop ... /> - property then object </ex:Foo>  <ex:Foo ...> - subject   <ex:prop> - property     <ex:Bar ... /> - subject (and implictly an object chaining from previous)  ... 

(rule of thumb: odd lines rdf:about, even lines, rdf:resource)

and using both rdf:about and rdf:resource on an element is almost always an error (you're either in a subject position or object position).

tl;dr

Avoid rdf:ID. Use rdf:about and rdf:resource much like an href, the former for subject, the latter for objects.

Additional

Forgot to mention that rdf:ID can be used on a property element, but it does something you may find unexpected: it reifies the triple. Avoid rdf:ID.

Answer 2

I would like to clarify a few of the excellent points in the provided answer, but only with respect to rdf:ID and rdf:about.

The tags are used to build a URI. If the full URI is not provided (such as rdf:ID="x"), then the generated URI is relative to the in-scope base URI usually derived from the document's location, but it can be specified with the xml:base attribute.

The point (mentioned above) is that rdf:about may be a fully qualified URI, so it is easy to just set it.

rdf:ID cannot be a fully qualified URI, but, you can still control that if you manually set the xml:base attribute.

The general rule of thumb for me, therefore, is to use rdf:about for a "globally known" identifier (when you want the URI to always be the same) and to use rdf:ID when describing a local resource whose URI is not cared about outside of the current document.