Re: Schema.org External Enumerations mechanism

On 25 April 2012 02:17, Guha <guha@google.com> wrote:
> After some discussion, etc., here is the next version of the proposal:
>
>  The canonical urls that Schema.org recommends for use will the urls for the
> entities on the reference sites (wikipedia, freebase, nist, etc.) When these
> reference sites add new entities (such as South Sudan as a new country),
> webmasters can immediately start using them.
>
>  In addition, to make the common use case much easier, Schema.org will
> provide documentation pages that list the entities (and their external
> urls), along with the caveat that the external entity is the primary source.
>
>  How does this sound?

I think this is heading in the right direction. I've updated
http://www.w3.org/wiki/WebSchemas/ExternalEnumerations accordingly,
with links into the wiki history for the earlier draft and for this
discussion.

My update is a first-pass and there may be (conceptual or technical)
remnants of the earlier design in there. We don't say anything about
URLs for countries vs pages about those countries, or whether we
expect HTML+Microdata (or RDFa, or whatever) at those URLs. I think
some flexibility is appropriate at this stage.

Daniel Dulitz notes (rightly) that the doc is currently a little
unclear as to whether we yet handle the externally enumerated subtypes
scenario. There are two syntax issues there: firstly that Microdata
doesn't support multiple types, and seems unlikely to change.
Secondly, when looking to the future and RDFa Lite usage, RDFa now
comes "out of the box" with some prefix bindings for existing W3C
vocabularies and for other commonly encountered general purpose RDF
vocabularies. I have started a conversation with Ivan (cc:'d) about
whether there might be scope to collaborate there.  This "RDFa initial
context" list, see http://www.w3.org/2011/rdfa-context/rdfa-1.1.html
and background doc http://www.w3.org/2010/02/rdfa/profile/data/ was
based on crawler stats from Yahoo and Sindice. It doesn't currently
include "bulk" vocabularies like e.g. DBpedia, but there might be
some scope for doing so. The syntactic benefit is that RDFa Lite's
'typeof' attribute can take a space-separated list of types:

So a simple description looks like this (from
http://www.w3.org/TR/rdfa-lite/#vocab--typeof--and-property )

<p vocab="http://schema.org/" resource="#manu" typeof="Person">
   My name is <span property="name">Manu Sporny</span>
   and you can give me a ring via
   <span property="telephone">1-800-555-0199</span>.
   <img property="image" src="http://manu.sporny.org/images/manu.png" />
</p>

Whereas if we wanted to say that Manu is in some other type too, eg. C
programmer, we can drop in http://en.wikipedia.org/wiki/Programmer or
(if the prefix is declared explicitly or via the W3C 'starter list'
just wiki:Programmer, ... or some more precise URL that means 'C
programmer'. Ideally Microdata would allow something like this too.

In RDFa then, sub-type external enumerations would be <p
vocab="http://schema.org/" resource="#manu" typeof="Person
wiki:Programmer"> ... how does that look?

Dan

Received on Wednesday, 25 April 2012 12:17:15 UTC