W3C home > Mailing lists > Public > public-data-shapes-wg@w3.org > January 2015

RE: Terminology: How to call "IRI or blank node"?

From: Irene Polikoff <irene@topquadrant.com>
Date: Fri, 9 Jan 2015 18:24:40 -0500
To: "'Peter F. Patel-Schneider'" <pfpschneider@gmail.com>, "'Holger Knublauch'" <holger@topquadrant.com>, <public-data-shapes-wg@w3.org>
Message-ID: <2bb601d02c63$72a82130$57f86390$@topquadrant.com>
Peter, how does one say in RDF that a string '5' is an instance of rdfs:Literal class?

-----Original Message-----
From: Peter F. Patel-Schneider [mailto:pfpschneider@gmail.com] 
Sent: Friday, January 09, 2015 6:13 PM
To: Holger Knublauch; public-data-shapes-wg@w3.org
Subject: Re: Terminology: How to call "IRI or blank node"?



On 01/09/2015 02:52 PM, Holger Knublauch wrote:
> On 1/9/15, 12:40 PM, Peter F. Patel-Schneider wrote:
>>
>> This is absolutely counter to the requirement that the working group 
>> must use rdfs:Resource in a way that abides by its meaning as defined 
>> in the RDF specification.
>
> Peter,
>
> I think the situation is not as clear-cut. Using rdfs:Resource in the 
> context of :valueType would IMHO be perfectly fine with the RDF specification.
>
> Look at rdfs:domain for example. The triple
>
>      ex:property rdfs:domain rdfs:Resource
>
> means that the domain of ex:property is IRIs or blank nodes- literals 
> are excluded because there is *another* rule in RDF which disallows 
> literals to appear as subjects in RDF triples.

This it not correct.

> All these semantic rules are conjoined.

That literals cannot appear as the subject of triples does not mean that literal values cannot be the first element of pairs in the extension of properties.  The rule that literals cannot appear as the subject of triples is a syntactic restriction, not a semantic one.

> In
> the same spirit, I see no reason why we cannot add an additional rule 
> that in the context of :valueType, the meaning is narrowed down to 
> IRIs or blank nodes. As long as we only narrow down the value space, 
> this is IMHO perfectly fine.

This is not fine at all, in my opinion.

> Another example is
>
>      ex:property rdfs:range rdfs:Literal
>
> According to the RDF Schema spec,
>
> "The class rdfs:Literal is the class of literal values such as strings 
> and integers. Property values such as textual strings are examples of RDF literals."
>
> Well, this is quite ambiguous. Does this mean that the following is in the range?
>
>      ex:MyLiteral a rdfs:Literal .
>      ex:MySubject ex:property ex:MyLiteral .


These three triples do not form an inconsistent RDF graph under the RDFS semantics.

> I guess not, because otherwise almost every existing RDF processor is wrong.

These processors may or may not be wrong, but that doesn't affect the definition of RDFS.

> However, how can rdfs:Literal then be a class, if it cannot really 
> have meaningful instances?

Why do you say that rdfs:Literal cannot be a class?  It has many meaningful instances, such as the string "5" and the integer 5.

> So apparently, rdfs:Literal is interpreted differently when it is used 
> in the context of rdfs:range.

Not at all.

> Finally, let's not forget that - in addition to remaining as 
> compatible as possible to the existing specs - this WG works in 
> unchartered territory in which many original RDF design decisions such 
> as the open world assumption are no longer valid. Furthermore, the 
> group also has the goal to create a user-friendly and intuitive 
> language. The fact that the vast majority of users find the current 
> interpretations of rdfs:Resource confusing should encourage us to think outside of the box.

The working group is free to set up new concepts if it feels that such are necessary.  The working group is not free to change the meaning of things that are defined in RDF.

> Regards,
> Holger

peter
Received on Friday, 9 January 2015 23:25:13 UTC

This archive was generated by hypermail 2.3.1 : Friday, 9 January 2015 23:25:14 UTC