W3C home > Mailing lists > Public > public-awwsw@w3.org > February 2011

Re: [Fwd: Reversing HTTP Range 14 and SemWeb Cool URIs decision]

From: Jonathan Rees <jar@creativecommons.org>
Date: Sun, 13 Feb 2011 13:40:07 -0500
Message-ID: <AANLkTi=xvFDaDJiJ+tQkOT20O9qxQp=sU_vg4HJ_XLkv@mail.gmail.com>
To: Harry Halpin <hhalpin@w3.org>
Cc: nathan@webr3.org, AWWSW TF <public-awwsw@w3.org>
On Thu, Jan 20, 2011 at 5:54 PM, Harry Halpin <hhalpin@w3.org> wrote:

> 6) The only practical example of code failing by lack of using it involves
> "owl:sameAs" and OWL reasoning, and owl:sameAs is demonstrably mis-used
> anyways (see paper in ISWC 2010) and OWL reasoning of any kind is not used
> often with Linked Data to begin with.

Harry, let me say that your messages have been tremendously useful in
helping me to understand the disconnect and I appreciate your
involvement in the conversation.

I've been turning this particular item (#6) over and over in my head
and I think I'm beginning to grok it. In OWL one almost never uses
sameAs directly, but you do use functional properties, which let you
infer identities (and detect nonsense, prove entailments, and so on).
If you eliminate inference, then term models are satisfying, and in
effect everything is a document (i.e. every term is interpretable as
data- or document-like). So I think you may be right - the LOD world
does not care about inference, so will probably never be on board with
any convention whose only purpose is inference.

This seems odd to me, since to me inference (both formal and informal)
is the whole value proposition for RDF. But I guess there are other
things to like about it.

If LOD properties were acknowledged to be data links instead of thing
links (this is what the acronym says, right?), we could then proceed
to document what the data means in terms of independently developed
theories of what the data is about, using OWL or some other form of
rational discourse. I'm beginning to like this. We know the meaning of
linked data will have to be reverse engineered in any case, just as
the XML, relational databases, and spreadsheets it comes from would
have to be in a logic-based approach, and this gives us a methodical
way to do it with a clean separation of concerns - LOD doing the
format conversion and dissemination, and others figuring out what it
means.

Jonathan
Received on Sunday, 13 February 2011 18:40:41 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Sunday, 13 February 2011 18:40:42 GMT