Re: @itemid and URL properties in schema.org

On Sat, Nov 5, 2011 at 7:43 AM, Gregg Kellogg <gregg@kellogg-assoc.com>wrote:

>
> I'll just note that this is a point that may require very special
> vocabulary-specific treatment in a processor generating RDF. Properties
> (@itemprop) create objects, not subjects. With some datatype information,
> such as we might include in a registry, we can know that a property using a
> literal value should be interpreted a URI reference. Assigning it as a
> subject is quite unnatural, as a BNode subject has already been created by
> the processor.
>

Realistically, I think there's always going to be some custom work
necessary in this area for schema.org processors.  No matter what we say,
webmasters are going to put the wrong URLs in the wrong slots and we're
going to have to do hacky things like try to determine whether the URL is a
canonical URL from the same site or a "foreign" URL to decide whether to
put it in the subject or object slot.

The schema.org group has been consistent in the statement that given the
choice of simplicity for webmaster vs. simplicity for processor, the
incremental implementation burden will *always* fall to the processor.

...


> >>    • Add Thing/sameAs for stating item equivalences (via URLs) across
> data sources/sites.
>
> Alternatively, schema:url could be a subProperty of owl:sameAs, which
> after a fair amount of reasoning can yield what you want (in RDF that is).
> If schema:url really is used to designate the subject, much better to use
> @itemid in the first place.


I get what you're saying and that argument is partly why there isn't a
sameAs property yet... there was enough confusion between the two models
(what I just proposed and what you're proposing) that it wasn't worth
including in the first round.

Honestly, though, I think this horse has left the barn.  If we were to try
to redifine 'url' at this stage I think it would create a big mess -- it's
already in fairly wide use.

So I still gravitate to the idea of @itemid and itemprop=url being presumed
to be local (subject) identifiers and itemprop=sameAs (conceptually similar
to owl:sameAs) being presumed to be foreign (object) identifiers.  (with
the acknowledgement that processors will likely still have to apply their
own heuristics as there's no way webmasters will get this right anywhere
near 100% of the time.)

-jason