W3C home > Mailing lists > Public > w3c-rdfcore-wg@w3.org > November 2001

Re: S3 or not S3

From: Sergey Melnik <melnik@db.stanford.edu>
Date: Fri, 02 Nov 2001 09:39:38 -0800
Message-ID: <3BE2DA5A.D2E1EDD5@db.stanford.edu>
To: jjc@hplb.hpl.hp.com
CC: w3c-rdfcore-wg@w3.org
Jeremy Carroll wrote:
> 
> Sergey:
> > <SUG3>: the interpretation of each literal symbol is fixed
> >        and is determined by its textual contents.
> 
> I am against.
> 
> I tend to agree with Peter, this just shifts the problem.
> 
> Let's think about it in terms of monotonicity.
> 
> In what I understand as Pat's approach, if I have a single triple
> 
> <foo> <bar> "lit" .
> 
> and no other information,
> then "lit" appears to be a string.
> 
> Somehow or other, I add typing information, and we have a non-monotonic
> change in our understanding of "lit", in that it is now interpreted as a
> particular typed value whose print-string is "lit".
> 
> If we are deeply committed to non-monotonicity we do a little somersault and
> say that the triple by itself has an object of unknown type whose
> print-string is "lit". Then adding the type information hasn't involve any
> change, just an augmentation of our knowledge!

Of course, as you point out, non-monotonicity is avoided above by
treating the literal as a bNode - something we know exist, but no other
information about this something is available (except its print-string).
What makes me uneasy about this approach is that it endorses the use of
literals instead of bNodes. In other words, I can write <foo> <bar>
"RDFCore WG" and assume that "RDFCore WG" denotes the RDFCore Working
Group. This intepretation would be valid if no further (schema)
constraints are provided. I'm not sure we want literals to represent by
default anything you like.

> This is parallel in Sergey's model:
> The triple
> <foo> <bar> _:bn .
> 
> represents the existence of something that is the bar of foo. But we don't
> know its type or value. Adding another triple
> 
> _:bn <type> "lit" .
> 
> tells us both the type and value.

Right.

> That this is just as much a somersault as before is shown by Peter's
> suggestion of there being a second (conflicting) type triple
> 
> _:bn <type2> "lit2" .
> 
> The contradiction arrived at is a sympton of an underlying non-monotonicity
> I think.

The above contradiction appears to me to have nothing to do with
non-monotonicity. We have two contradicting definitions (not a
definition that is overridden by another definition). There is no valid
interpretation that satisfies all three statements above, right?

> 
> Jeremy

Sergey
Received on Friday, 2 November 2001 12:26:05 EST

This archive was generated by hypermail pre-2.1.9 : Wednesday, 3 September 2003 09:42:28 EDT