[Top] [All Lists]

Re: [ontolog-forum] {Disarmed} Re: OWL and lack of identifiers

To: Waclaw Kusnierczyk <Waclaw.Marcin.Kusnierczyk@xxxxxxxxxxx>
Cc: Ontolog Forum <ontolog-forum@xxxxxxxxxxxxxxxx>
From: Pat Hayes <phayes@xxxxxxx>
Date: Mon, 16 Apr 2007 14:18:47 -0500
Message-id: <p06230904c2497855a0b1@[]>
>Pat Hayes wrote:
>>>Pat Hayes wrote:
>>>>>Re: Peter F Brown's post (Sat, 14 Apr 2007 09:35:14)
>>>>>Peter writes:
>>>>>The spec is clear, yesS. but an object is not the same thing as the
>>>>>address of the object - (according to the RFC, I *am* my address): the
>>>>>object needs identity as much as the address of it does. That is where I
>>>>>feel this axiom of the W3C falls downS
>>>>>Clearly there is a problem here.  But we should be careful to
>>>>>distinguish confusedly designed frameworks from confused documentation
>>>>>of well-designed ones.
>>>>>While RDF specifications, for example, are relatively clear and sound,
>>>>>the RDF primer provides an abundance of examples such as:
>>>>>ex:index.html  exterms:creation-date  "August 16, 1999" .
>>>>>ex:index.html  dc:language            "en" .
>>>>>supposed to state that "August 16, 1999" is the creation date of a page
>>>>>and "en" is he language of a page, while both are literal strings and
>>>>>*not* identifiers for a date and a language, respectively.
>>>>What??  Why should a string not be an identifier? In fact, it 
>>>>seems to me that *all* identifiers are strings. And the second 
>>>>example uses a language tag which is taken from an Internet 
>>>>standard for language tagging: what could be a better example of 
>>>>an agreed identifier? Why is this confused?
>>>Hold on.  It is not whether something is a string or not which 
>>>counts, but how it is to be interpreted.  Of course, "August 16" 
>>>can be an identifier for anything you may wish.  But it is a 
>>>string, not a date.
>>It *is* a string, of course. It *identifies* a date. You said it 
>>was not an identifier, not that it was not a date.
>Yes, and I said that because this follows from RDF specs.  A literal 
>string appearing as the object of a triple is treated as a literal, 
>i.e, as a self-referential entity, rather than an identifier for 
>some other entity.  In this context, "August 16" is not an 
>identifier for a date, it is an identifier for itself.  Is this 
>better?    (01)

In the RDF model theory, it denotes itself. But it may also be an 
identifier for something else, using a system of identification 
external to RDF.    (02)

>>>But as the object of a triple, "August 16" is a literal, not a 
>>>URI, and in RDF, a literal is (supposed to be) self-referential.
>>Plain literals, yes. It would indeed be better practice to have the 
>>object of that triple be a typed literal. But this is an example 
>>from the *primer*; one step at a time.
>>>  Thus, the denotation of "August 16" is "August 16", not a date.  Thus,
>>><u1> <u2> "August 16"
>>>effectively means that the denotation of u1 is in the denotation 
>>>of u2-relation with "August 16" (the string), not with a date.
>>But that string can be interpreted as denoting a date, using widely 
>>used conventions for denoting dates. So this is a perfectly valid 
>>RDF extension, speaking strictly, where 'extra' semantic conditions 
>>are imposed (in this case, that the date-string is interpreted 
>>conventionally). We anticipated that RDF would be widely used with 
>>such conventions added.
>Can't see how this would be a good practice;  for a dumb machine, 
>you leave the question:  is "August 16" supposed to identify a date 
>(or whatever else), or is it supposed to identify "August 16"?    (03)

The RDF machine, which is unusually dumb, can't tell. (It also 
doesn't know that owl:sameAs means anything in particular, for 
example.) But other entities using the formalism - some of which may 
be machines which know other conventions - may. RDF isn't supposed to 
be an AI Knowledge Representation language, standing alone and 
supporting all notions of meaning. Its primary function is to act as 
a means of transporting meaning rather than representing it. It would 
be normal for RDF to convey some relationships using 'signs' to which 
it - RDF - assigns no particular meaning. This is why we said that 
plain literals denote themselves, rather than denoting an xsd string, 
say: to ensure that no matter how much other specs (such as XML 
Schema part 2) get re-written or amended, the exactness of a semantic 
transfer using RDF will not be impaired.    (04)

>While in this case you might use the range of the relation as a 
>guidance, it may not always be so easy.  Consider the case:
><text-uri> <contains-relation-uri> "the text 'hello'"
>Now, is "the text 'hello'" (replace single quotes with escaped 
>double quotes if preferable) an identifier for "the text 'hello'" or 
>for "hello"?  It won't help that (the referent of) 
>contains-relation-uri has the type 'string' as its range.
>My point is, it seems far better to say that a literal is 
>self-referential rather than that it may or may not be, depending on 
>some unspecified context.    (05)

It is self-referential in pure RDF. But it may still identify 
something else in a larger system of interpretation. RDF is not 
intended to be incompatible with such larger uses: quite the contrary.    (06)

Pat    (07)

>>>>>   But from
>>>>>this apparently confused example one should not infer that RDF itself is
>>>>>confused (in this respect).
>>>>>(This is not to say that RDF is not confused.)
>>>>One can make many valid criticisms of RDF, but being confused 
>>>>isn't one of them.
>>>I did not say that RDF was confused either.  I just said I hadn't 
>>>said it were.
>>>>>Wacek Kusnierczyk
>>>>>Department of Information and Computer Science (IDI)
>>>>>Norwegian University of Science and Technology (NTNU)
>>>>>Sem Saelandsv. 7-9
>>>>>7027 Trondheim
>>>>>tel.   0047 73591875
>>>>>fax    0047 73594466
>>>>>Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/ 
>>>>>Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
>>>>>Shared Files: http://ontolog.cim3.net/file/
>>>>>Community Wiki: http://ontolog.cim3.net/wiki/
>>>>>To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx
>>>Wacek Kusnierczyk
>>>Department of Information and Computer Science (IDI)
>>>Norwegian University of Science and Technology (NTNU)
>>>Sem Saelandsv. 7-9
>>>7027 Trondheim
>>>tel.   0047 73591875
>>>fax    0047 73594466
>Wacek Kusnierczyk
>Department of Information and Computer Science (IDI)
>Norwegian University of Science and Technology (NTNU)
>Sem Saelandsv. 7-9
>7027 Trondheim
>tel.   0047 73591875
>fax    0047 73594466
>------------------------------------------------------    (08)

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
phayesAT-SIGNihmc.us       http://www.ihmc.us/users/phayes    (09)

Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/  
Subscribe/Config: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/  
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/ 
To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx    (010)

<Prev in Thread] Current Thread [Next in Thread>