W3C home > Mailing lists > Public > www-tag@w3.org > July 2007

Re: Terminology Question concerning Web Architecture and Linked Data

From: Jeremy Carroll <jjc@hpl.hp.com>
Date: Tue, 24 Jul 2007 10:46:50 +0100
Message-ID: <46A5CA8A.2040301@hpl.hp.com>
To: Mark Baker <distobj@acm.org>
CC: "Booth, David (HP Software - Boston)" <dbooth@hp.com>, Dan Brickley <danbri@danbri.org>, Chris Bizer <chris@bizer.de>, www-tag@w3.org, semantic-web@w3.org, Linking Open Data <linking-open-data@simile.mit.edu>

Mark Baker wrote:
 > Any of them, it doesn't matter for the purposes of my argument.  So if
 > there's eight URIs in total, I claim that none of them directly
 > identify the same resource.

Are you in essence pointing to underlying philosophical myths that we 
use in everyday reasoning.

Reasoning about identity and non-identity of things is a skill that we 
all use to survive.

However, it is largely ill-founded, and closer examination makes it hard 
to defend (except on evolutionary grounds: we're alive, our reasoning 
processes are effective - but not necessarily true)

The underlying notion of resource presupposes that identity is a 
non-problematic concept, whereas say the Tim Berners-Lee who invented 
the web, and the person who we currently think of as Tim Berners-Lee 
have fairly little in common - someone better versed than me in biology 
could estimate the number of atoms that they share, for instance. The 
Tim Berners-Lee who invented the Web was an unknown, the person we 
currently think of as Tim Berners-Lee is not. Many other attributes will 
have changed.

This is not merely a temporal argument.

For many of us, a key feature of TimBL is that he invented the Web. I 
suspect that for those closest to him, this is not a key feature. So 
that different descriptions of him by different people will emphasize 
different attributes. And once we have that difference in emphasis, we 
have enough for the philosophers, or novelists, to get their teeth into.

 From an engineering point of view we probably want to avoid such 
issues, which may involve following 'common sense' despite some 
plausible reservations, and I feel that in general, Chris Bizer's 
approach does this effectively.

Jeremy

PS I am wondering about the impact of polyjuice potion on this discussion :)






Mark Baker wrote:
> 
> On 7/23/07, Booth, David (HP Software - Boston) <dbooth@hp.com> wrote:
>> > From: Mark Baker
>> >
>> > . . . I *fully* agree
>> > that the representations returned from all four of those URIs
>> > are *about* Tim, I just don't believe that makes them the same
>> > resource.
>>
>> This is confusing for a couple of reasons.  First, because each URI for
>> TimBL (a non-information resource) involves a secondary URI for an
>> information resource that serves a declaration of the first URI.  So it
>> isn't clear which URIs you're talking about, as referring to "the same
>> resource".
> 
> Any of them, it doesn't matter for the purposes of my argument.  So if
> there's eight URIs in total, I claim that none of them directly
> identify the same resource.
> 
> To elaborate further, I'm not saying that the publishers of those URIs
> cannot assert "This URI (directly) identifies Tim".  Of course they
> can.  Let's even assume they've done so.  But now let's step back and
> look at those assertions in the aggregate, as an independent observer.
> Do they all directly identify the same resource?  Well, if they did,
> then they would all be aliases (the word Chris used in his initial
> question) such that each URI would be able to be substituted for any
> of the others in all contexts.  Right?  Consider though, that URI #1
> could also be used to indirectly identify "An example dbpedia.org
> page", or "A document containing the string 'Depiction of Tim
> Berners-Lee", but you couldn't say the same thing about any of the
> other URIs.  Therefore they are not aliases and so do not all directly
> identify the same resource.
> 
> I suppose that's just a long-winded way of pointing out that each of
> us means something different by "Tim" (even the "independent
> observer").  We all have the same physical person in mind, but we each
> introduce our own biases in the information we publish about him, and
> that's what makes our URIs directly identify different resources.
> 
> On the upside, all those URIs can still be used to *indirectly* identify 
> Tim.
> 
> FWIW, I'm not certain if that explanation is consistent with REST or
> not.  It's not directly covered by the dissertation AFAICT, nor have I
> heard Roy say anything about how agency affects direct and indirect
> identification.  But it makes sense to me based on my understanding of
> REST and my experiences applying it.  I hope Roy will correct me if
> I'm wrong.
> 
> Mark.

-- 
Hewlett-Packard Limited
registered Office: Cain Road, Bracknell, Berks RG12 1HN
Registered No: 690597 England
Received on Tuesday, 24 July 2007 09:47:15 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Thursday, 26 April 2012 12:47:46 GMT