W3C home > Mailing lists > Public > www-rdf-comments@w3.org > October to December 2004

Re: W3C specification error

From: Andrea Proli <aprol@tin.it>
Date: Tue, 30 Nov 2004 00:01:31 +0100
Message-ID: <006e01c4d667$5dc650f0$0201a8c0@lycantro>
To: "Pat Hayes" <phayes@ihmc.us>
Cc: "Dan Brickley" <danbri@w3.org>, <fmanola@acm.org>, <www-rdf-comments@w3.org>

Thank you all for your feedback.
Although I hardly can understand what are you talking about
(when I raised this issue I was reading the specification for the
first time and never worked with RDF before), I really appreciated
the promptness of your response.

But please let me flow through a very simple reasoning (well, I know
you're right, this is not a polemic: I just hope that making more explicit
my erroneous beliefs will give you the opportunity to better explain me
where exactly my mistake is).

If a "statement" is the connection between a "subject" and an "object"
that consists in that "subject" having that "object" as the value of the
connecting "property", why shouldn't that "property" really be modeled
as an "rdf:Property", instead of simply being an "rdfs:Resource"?
Clearly, using "rdfs:Resource" allows you to model illegal statements,
but why should this be desirable? Why should illegal statements
exist? As to my point of view, a parser should detect it and raise
an error whenever an illegal statement is made... So, if my rationale is
wrong (and I really know it is, but it's difficult to me to realize why,
and that is certainly due to my ignorance), why the same does not apply
to the "type" property, where the "rdfs:range" is constraint to "rdfs:Class"
instead of being "rdfs:Resource"? Shouldn't be equally desirable to
model that something is an instance of something else that is not an
"rdfs:Class", if an "rdfs:predicate" can have an object  that is not
an "rdf:Property" (in both cases, due to an error)?

I know the following could seem a little stupid to you, but in my opinion 
letting
the "rdfs:range" of "rdfs:predicate" being a not-better-specified 
"rdfs:Resource"
just because reifications should be able to model statement errors is pretty 
like
letting horses having "things" instead of "feet", just because someone could
erroneously state that a horse has four "frogs" attached to its legs - put 
it simple,
it's conceptually incorrect...

Can you please tell me where exactly is my mistake?

Ok, that's all, I won't bother you anymore, I promise!
Thanks again for your great feedback,

Andrea

----- Original Message ----- 
From: "Pat Hayes" <phayes@ihmc.us>
To: "Dan Brickley" <danbri@w3.org>; "Andrea Proli" <aprol@tin.it>
Cc: <www-rdf-comments@w3.org>
Sent: Monday, November 29, 2004 9:14 PM
Subject: Re: W3C specification error


> >Thanks, I'll take a look. I remember (dimly!) the group making some 
> >decision on
>>this which seemed counter-intuitive. I've copied www-rdf-comments to put
>>your note on the record, hope that's OK. I think what happened might be
>>that the mathematics of having the more constrained form were quite
>>tricky, so we ended up saying just 'Resource'...
>
> I also dimly recall this decision being made, and that at the time the 
> reasons seemed good to me. I don't think it was because the narrower 
> interpretation presented any particular mathematical difficulties. It may 
> have been the observation that reification should be able to describe 
> 'illegal' RDF, in which a non-property URIreference is used in a predicate 
> position in a triple. At any rate, the mere possibility of such an error 
> occurring means that one should not be able to conclude, merely from the 
> fact of a URI being used in some RDF in this position, that it really 
> does, in fact, denote a genuine property (which would be the effect of 
> having the range be rdf:Property)
>
> It might be worth remarking that to have rdfs:Resouce as a domain or range 
> is never an error, since in RDFS domains and ranges can be conjoined. It 
> is more like a kind of resignation: one is saying that the subject or 
> object of the property may be anything whatsoever, unless of course 
> further information is supplied which restricts them in some other way or 
> for some other reason. One can see this by looking at the RDFS inference 
> rules (http://www.w3.org/TR/2004/REC-rdf-mt-20040210/#RDFSRules), where 
> rdfs3 allows you to conclude in this case that the type of the object of 
> any assertion of rdf:Property must be rdfs:Resource; but one knew that 
> already, from rdfs4b. So this 'vacuous' range only provides some redundant 
> information.
>
> Pat Hayes
>
>>Thanks again,
>>
>>Dan
>>
>>* Andrea Proli <aprol@tin.it> [2004-11-29 02:37+0100]
>>>  Dear Mr. Brickley,
>>>  I think I have found an error in the W3C Specification "RDF Vocabulary 
>>> Description Language 1.0: RDF Schema" (10 February 2004).
>>>  Since you are the co-editor of the above mentioned specification, I 
>>> thought it could be useful for you to receive this notification.
>>>  However, I am new to RDF, so please forgive me in the case I'm wrong.
>>>
>>>  In the very last paragraph of Section "5.3.3 rdf:predicate" you wrote: 
>>> "The rdfs:domain of rdf:predicate is rdf:Statement and the rdfs:range
>>>  is rdfs:Resource". I found it inconsistent with previous definitions, 
>>> and I argue that the rdfs:range of rdf:predicate should be rdf:Property
>>>  instead of rdfs:Resource.
>>>
>>>  Obviously, I wrote to you because I didn't find this error in the 
>>> errata section. Am I wrong about it?
>>>  Thank you for your attention,
>>>
>>>  Andrea Proli
>
>
> -- 
> ---------------------------------------------------------------------
> IHMC (850)434 8903 or (650)494 3973   home
> 40 South Alcaniz St. (850)202 4416   office
> Pensacola (850)202 4440   fax
> FL 32502 (850)291 0667    cell
> phayes@ihmc.us       http://www.ihmc.us/users/phayes
> 
Received on Tuesday, 30 November 2004 13:37:38 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 5 February 2014 23:40:22 UTC