Re: Meaning of property "url"

Hi Dan,
If I understand your post correctly, it would not be correct to use for pointing from an abstract object (e.g. a product) to the canonical URL of its Web page, or from a person to its main homepage?

We are using foaf:page in many GoodRelations patterns, so it would be crucial to know whether it is correct to use for

a) authoritative identifiers for entities (like or an identifier (!) from Freebase)
b) HTML pages and PDF documents containing descriptions of the objects (like

or both.

My current understanding was that 

1. can be used for both and that search engines will try to guess the nature of the URI/URL, so you can use it for related pages (and PDFs?) as well as for authoritative Web-scale identifiers, and that

2. itemid in Microdata and about in RDFa can be used for assigning identifiers.

But now you say that itemid and are "broadly equivalent", which confuses me.

If that is meant as I read it, then we likely need to add foaf:page and foaf:homePage to

Thanks in advance for clarifying this.


On Oct 23, 2012, at 5:25 PM, Dan Brickley wrote:

> A few observations -
> * We want all Things to be identifiable using URI/IRI/URL identifier notations
> * In the HTML5 the specs use 'URL" rather than "URI"; at launch
>'s primary notation was HTML5 MIcrodata, so we inherit that
> usage. But also targets mainstream developers and
> publishers who often are not so familiar with 'URI' or 'IRI', but feel
> they know what an 'URL' is
> See
> HTML5's usage of 'URL' is explicitly in terms of URI and IRI and the
> notion of a 'resource'
> * In RDF, we use 'resource' as a synonym for Thing (ie. all things can
> be considered resources), rather than something like
> 'http-accessible-information-object', which seems to be some people's
> reading of the term. I think is closer to the 'thing'
> reading.
> * Microdata has an 'itemid' attribute, for Thing identifiers
> (analagous to resource= in RDFa Lite), see
> for the RDFa version.
> * The use of is broadly equivalent to using itemid or
> resource attributes from Microdata and RDFa; it takes the items Web
> identifier and expresses it explicitly as a property value.
> * There is plenty of room still for interpretation, best practice,
> clearer guidelines; the "How do we identify real world entities"
> debate is as old as the Web.
> *'s deployment in mainstream Web content places some
> practical constraints for some publishers; for example, a Movie site
> where a page /person_321/ links to /tvshow-67241/ might have an
> itemprop="actor" (microdata) or property="actor" (rdfa) annotation.
> That's nice and simple, but parsing gives the actor TV show the same
> URI/IRI/URL as the page describing them; i.e. the http-range-14.  A
> more complex site design (markup and identiifers) that gives different
> IDs to pages and entities is of course possible, but it's not clear
> we'd see strong adoption easily.
> So I'd not read too much into 'url'. It's somewhere you can put a Web
> identifier for the thing being described. As conventions for this in
> the Web standards community mature, we should be able to be more
> precise on this.
> Dan
> p.s. is somewhat related

martin hepp
e-business & web science research group
universitaet der bundeswehr muenchen

phone:   +49-(0)89-6004-4217
fax:     +49-(0)89-6004-4620
www: (group) (personal)
skype:   mfhepp 
twitter: mfhepp

Check out GoodRelations for E-Commerce on the Web of Linked Data!
* Project Main Page:

Received on Wednesday, 24 October 2012 10:02:43 UTC