W3C home > Mailing lists > Public > public-vocabs@w3.org > June 2012

Re: schema.org/sameThingAs proposal

From: Joshua Shinavier <josh@fortytwo.net>
Date: Fri, 22 Jun 2012 19:11:19 -0400
Message-ID: <CAPKNUSvcQPUhpQZiM5J9htHtHg06cuUnqXd=OCzs40KnFroCLg@mail.gmail.com>
To: Dan Brickley <danbri@danbri.org>
Cc: public-vocabs@w3.org
On Fri, Jun 22, 2012 at 4:47 PM, Dan Brickley <danbri@danbri.org> wrote:
> On 22 June 2012 21:52, Joshua Shinavier <josh@fortytwo.net> wrote:
>> Hi Dan,
>> I'm an instant fan of this proposal.  sameThingAs would be an obvious
>> win for Semantic Web interoperability.  Just a trivial
>> question/comment about the name: why not call it "sameAs"?  Since it's
>> a property of Thing and the expected type is also Thing, the "Thing"
>> in the name seems redundant.  And "sameAs" is shorter...
> Thanks. Care to show support in public? :) I don't think the design is
> quite there yet, but something less brittle / strict than owl:sameAs
> would be useful.
> People often use 'sameAs' as shorthand for owl:sameAs; I think two
> properties with the same name but different meaning could be quite
> confusing.

A single property which is used with multiple meanings [1] is
*already* confusing, but I can see the sense in not compounding the
confusion if it can be avoided.

> We have done this on a very limited scale with schema:range and
> :domain, but purely within our own documentation. The sameThingAs
> property would be much more widely deployed, so calling it 'sameAs'
> could be pretty confusing for people used to the owl version.
> Also imho interposing 'Thing' in the middle makes the intent clearer;
> the idea is that we're focussed on a single entity, rather than some
> vaguer notion of similarity. This is a point that has led people to
> use owl:sameAs without realising what it really means, ie. that it's
> for use in situations when there really is just one entity not two.
> With sameThingAs we emphasise that, but also (yes, confusingly) allow
> document ids to be in the mix too...

I probably need to read the proposal again, but it's not clear to me
why sameThingAs needs to be able to link a *thing* with a *document
about a thing*, directly.  In the url example, it's pretty clear that
Alice, the subject of the Person markup, is a person, while alice.html
is the relative URL of a page about Alice.  When would it be useful to
point from Alice to the URL of a page about Alice using sameThingAs,
as opposed to pointing to another Thing which links to the page using
the "url" property?  Am I making going in circles yet?  To express the
fact that two web pages are about Douglas Adams, why not use markup
like this, which clearly distinguishes things from documents:

<div itemscope itemtype="http://schema.org/Person">
  <a href="http://en.wikipedia.org/wiki/Douglas_Adams"
itemprop="url">Douglas Adams</a> in Wikipedia is the same as
  <span itemprop="sameThingAs" itemscope itemtype="http://schema.org/Person">
    <a href="http://www.rottentomatoes.com/celebrity/douglas_adams/"
itemprop="url">Douglas Adams</a> in Rotten Tomatoes</span>


[1] http://www.w3.org/2009/12/rdf-ws/papers/ws21

> Tricky stuff!
> Dan
Received on Friday, 22 June 2012 23:15:52 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:29:23 UTC