Literals representing people? (was: Re: relationship of rdfs:Literal to rdfs:Resource)

Hi,

Garret Wilson wrote:
> Referring again to 
> http://www.w3.org/TR/1999/REC-rdf-syntax-19990222/#basic , are you 
> proposing that Ora Lassila (who should know, being the editor of the 
> first RDF specification) claimed that the web page 
> http://www.w3.org/Home/Lassila was created by the character string 
> O-r-a-_-L-a-s-s-i-l-a? (Have the semantics of RDF literals have changed 
> a lot since 1999?)

I would *hope* that "Creator," in the first example, was to be taken to 
be interpreted as "the name of the creator as a character string" ;-)

But I may be wrong-- I think it would be good if some other members of 
the list, in particular members of the core WG, could speak up on how 
literals are to be interpreted in cases like this.

>>     Literals are used to identify values such as numbers and dates by
>>     means of a lexical representation.
>>
>> and I don't think that "values such as numbers and dates" includes 
>> people.
> 
> I would take "such as..." to denote a subset, meaning the set containing 
> numbers and dates but not necessarily exclusive to people.
 >
 > Surely you're not asserting that literals can *only* denote numbers and
 > dates---under that definition, a literal could not represent a character
 > string.

A subset: yes, of course, but I take "values" and the particular 
examples given as indicating the characteristic that these represent 
some kind of atomic "data"-- numbers, dates, strings, byte sequences...

Apparently this deserves a comment to RDF Core, since we interpret their 
current spec differently on this point.

>> Besides, if you allowed a literal "Garret Wilson" to identify a 
>> person, you would have a *real* problem: What if *I* used it to denote 
>> a *different* person?
> 
> We already have this problem---what if I used "999" to refer to the 
> integer value 1000-1, and *you* used 999 to refer to the character 
> string 9-9-9? Then Ora Lassila comes along and uses "999" to refer to a 
> date, defined as the number of seconds past 22 February 1999 (sorry, 
> Ora). ;)

I would take the view that a plain literal always denotes a (character 
string, language tag) pair and that properties that can take plain 
literals need to be defined in a way that makes this a reasonable 
interpretation. Typed literals can be used to refer to things like integers.

> But maybe I'm wrong and literals are only supposed to be able to 
> represent logical number-like things. If so, maybe it could be made a 
> little clearer in one of the specs.

Or I'm wrong and literals are supposed to be able to represent arbitrary 
resources like people. If so, maybe it could be made a little clearer in 
one of the specs. ;) ;)

In particular, it should then be clear how to deal with the problem that 
the same literal can refer to more than one resource.

> This discussion (while fun) is all tangential to the main point: I want 
> to identify my resources by reference URIs (I'll even take node IDs), 
> even if I've referred to them using some lexical notation.

Not completely IMHO: If a literal can identify more than one resource, 
and you can state somehow that a literal and a URI node represent the 
same thing, then you can get the following situation. Graph 1::

     ex:foo   owl:sameAs   "Garret" .
     ex:foo   owl:sameAs   "Garret Wilson" .
     ex:foo   owl:fullName "Garret Wilson" .
     ex:foo   foaf:mailbox <mailto:garret@globalmentor.com>.

Graph 2::

     ex2:bar  owl:sameAs   "Garret" .
     ex2:bar  foaf:homepage <http://www.mrao.cam.ac.uk/~garret/>.
     ex2:bar  ex2:addr     "Cavendish Laboratory, Madingley Road
                            Cambridge CB3 0HE, United Kingdom" .

Graph 3::

     ex3:baz  owl:sameAs    "Garret" .
     ex3:baz  ex3:firstName "Garret" .
     ex3:baz  ex3:lastName  "Sueng" .
     ex3:baz  foaf:homepage <http://pages.cpsc.ucalgary.ca/~sueng/>.
     ex3:baz  ex3:employer  <http://pages.cpsc.ucalgary.ca/rdf#UCalgary>.

(``owl:sameAs`` could be replaced with any future RDF mechanism to 
designate equality between a literal and a URI-identified resource.)

Taking these-- true-- graphs together, we find that "Garret" is "Garret 
Wilson," a person with first name Garret, last name Sueng, full name 
Garret Wilson, with an address in Cambridge, UK, employed by University 
of Calgary, Canada, with two home pages and an email address at 
globalmentor.com.

Your view of literals *does* raise this concern, so discussion of how to 
deal with it does seem appropriate. ;-) Do you have anything in mind 
about the above?

Cheers,
- Benja

Received on Sunday, 28 September 2003 07:26:26 UTC