W3C home > Mailing lists > Public > www-rdf-interest@w3.org > September 2004

RE: web proper names

From: Jon Hanna <jon@hackcraft.net>
Date: Mon, 20 Sep 2004 21:58:49 +0100
To: <www-rdf-interest@w3.org>
Message-ID: <008d01c49f54$a10d01f0$0201a8c0@Lugh>

This actually inline 
> with the TAG (as it stands),

The TAG stands with httpRange-14 undecided.

 and solves the problem point 
> blank without either a new URI scheme or new RDF predicates. 

New RDF predicates are 2-a-penny. You have to try harder to beat a few
new RDF predicates as an easy solution.

I'll preface the following by noting that
http://purl.org/dc/elements/1.1/creator has already existing in its
definition which impinges on it's use in the examples. The only
recommendation (as opposed to proposed recommendation) which uses DC
properties as RDF predicates only uses creator in statements with
literal objects (the name of the creator). Might I suggest
http://xmlns.com/foaf/0.1/maker as a predicate with similar meaning but
clearer semantics.

> One could just say:
> 
> http://www.w3.org/People/thompson/
> http://purl.org/dc/elements/1.1/creator
> http://www.ltg.ed.ac.uk/~ht/HenryThompson.wpn
> 
> And http://www.ltg.ed.ac.uk/~ht/HenryThompson.wpn returns a 
> representation inline with the RDDL document specified by the 
> WPN paper. 

That's assuming that you accept that
http://www.ltg.ed.ac.uk/~ht/HenryThompson.wpn can identify a person (not
universally accepted, though I personally agree).

I don't see the point in only having an RDDL representation (why not
have an RDF representation too). Right now we have this hypothetical URI
being used in hypothetical RDF triples. Get some hypothetical RDF there
first, worry about hypothetical RDDL later!

> Comments in-line on other threads!
> 
> > The analogy is false IMO. Mentioning a URI is saying something like
> > "The  URI 'http://www.example.net/blah' has 27 characters, 
> uses the common
> > convention of naming a webserver 'www', has no query-string 
> and doesn't
> > > use any of character escapes defined in RFC 2396".
> 
> 	Correct in one sense, wrong in the other. In the sense that
> you are mentioning a URI without any reference to the resource it 
> identifies, correct. However, the trouble appears 
> that URIs are meant to *identify* resources. You can use a 
> URI to refer 
> to either the representation it can retrieve (*mention*, as in the 
> word 'rice' in a sentece could be the representation of a 
> bowl of rice on 
> my table) or use the URI to refer to a thing itself (*use*, as in 
> directly talking about the bowl of rice on my table). I don't see
> anywhere in the specs where it says that's impossible - 
> indeed, RDF,TAG
> and the new URI draft encourage this behavior.

I agree that we do need to mention a URI to talk about a particular
representation that can be retrieved by mentioning it (we also need to
mention the values of any headers that were mentioned in the Vary header
of the response, I didn't spot anything about that in your paper). I
disagree that there is any mentioning going on in your examples which
weren't making incorrect statements.

> Jon's short explanation on the TAG mailing list dodges the question.
> 
> What people actually do is this:
> <http://www.w3.org/People/thompson/> 
> <http://purl.org/dc/elements/1.1/creator>
> "Henry Thompson" 
> 
> But you cannot make many meaningful statements about the string "Henry
> Thompson" since it doesn't have a URI.

You can make plenty of meaningful statements containing the string
"Henry Thompson". I agree though that in general a person is a more
useful object than a name can we switch to using foaf:maker in our
examples?

 Thus, it would 
> actually make sense
> to use:
> 
> <http://www.w3.org/People/thompson/>
> <http://purl.org/dc/elements/1.1/creator>
> <http://www.ltg.ed.ac.uk/~ht>
> 
> But then *a machine* doesn't know whose web-page is talking 
> about which 
> man, or if it's two web-pages, or two men.

A machine doesn't know that there are any pages or any men. However in
the following:

<http://www.hackcraft.net/foaf/>
<http://www.w3.org/1999/02/22-rdf-syntax-ns#type>
<http://xmlns.com/foaf/0.1/Document> .
<http://www.hackcraft.net/foaf/> <http://xmlns.com/foaf/0.1/maker>
<http://www.hackcraft.net/jon/> .
<http://www.hackcraft.net/jon/>
<http://www.w3.org/1999/02/22-rdf-syntax-ns#type>
<http://xmlns.com/foaf/0.1/Person> .

Makes it clear to a machine that <http://www.hackcraft.net/foaf/> is a
<http://xmlns.com/foaf/0.1/Document> (whatever that is) and
<http://www.hackcraft.net/jon/> is a <http://xmlns.com/foaf/0.1/Person>
(whatever that is). Not only can machines work that out, but they act
upon RDF classes - that is they can "know" (well enough to do useful
work) what a foaf:Document and what a foaf:Person is. More to the point,
there is software out there already that does.
	
> Still, Web Proper Names are meant to solve the 
> problem in a RDF-neutral matter, and can solve via either a new URI
> scheme or using a new format (RDDL-based) for representing things qua
> things. Enjoying the discussion...

This bit boggles me. Why do we want to be RDF-neutral?

If you can so how it can be done using HTML then I'd say "sure, we don't
need a new technology". If you can so how RDF can't do it I'd say "sure,
we need a new technology". As it is you're ignoring a technology in
place which works close to the deepest roots of web architecture (URIs)
and inventing a new one. Or a new URI scheme; which in practice means
either using a new technology (that can dereference or otherwise act on
the URIs) or using the URIs thereof in RDF in which it won't matter two
hoots what the scheme is.
Received on Monday, 20 September 2004 20:59:04 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:52:09 GMT