Re: [Fwd: [Moderator Action] RE: RDF Semantics: Interpretations and Modelling]

>-----Forwarded Message-----
>
>From: Ossi Nykänen <onykane@butler.cc.tut.fi>
>To: pat hayes <phayes@ai.uwf.edu>
>Cc: Massimo Marchiori <massimo@w3.org>, w3c-rdfcore-wg@w3.org
>Subject: [Moderator Action] RE: RDF Semantics: Interpretations and Modelling
>Date: 04 Feb 2003 04:40:18 -0500
>
>
>
>Dear Patrick Hayes,
>
>Thank you for the clarification.
>
>>  > This seems
>>  > to nullify the work and efforts spend on defining,
>>  > e.g., the rdf:bag container
>>
>>  What work and effort? What definition?
>
>This is my opinion only but if RDF Semantics (note: as a design choice) is
>not willing to differentiate, e.g., bags and alts in entailments, then
>perhaps the model-theoretic approach isn't suitable for the purpose.

The issue is not whether or not it is 'willing' to make such a 
differentiation, but whether it is *possible*.

It seems to me that this is a non-issue, because RDF does 
differentiate bags and alts  in entailments. It may seem trivial, but 
the distinction is precisely that a bag is something in the class 
rdf:Bag, and an alt is something in the class rdf:Alt.  That is, RDF 
enables one to make the distinction, assert that something is one or 
the other, and moreover to tell you all the information you usually 
need to know about it, viz, what is in it. Of course RDF does not 
*implement* bags and alts; it does not implement anything in this 
sense. It is a descriptive language for describing things, not for 
implementing them.

It is true that the description of what is in a container is given in 
the same way for bags, seqs and alts: but so what? As long as you 
know that something is a bag (say), and you know what the things in 
it are, why does it matter if the list of things in it was given to 
you in some order? You are free to ignore that order if you wish to, 
just as you are free to ignore any piece of information that does not 
interest you or you know to be irrelevant. The ordering of the 
description need not be relevant to any ordering of the thing being 
described: and in the case of bags, we know that it is not relevant. 
(In an earlier discussion of this point the following analogy came 
up. Imagine a police report describing a body, which lists the items 
found in the victim's trouser pocket. The list is of course ordered, 
being a list, but that does not mean that the pocket was ordered.)

>My comment is purely practical. If differentiating bags and alts is
>considered important in RDF, this makes you wonder the role of RDF
>Semantics in applications of RDF (*). In practice, it would seem to be
>safer to use, e.g., (now imaginary) ibm-glossary:Bag and ibm-glossay:Alt
>with (perhaps rather restricting but) fixed entailment semantics  given by
>IBM

Please suggest what those entailments might be, and what uses they 
might have. If they make sense I am quite willing to suggest that 
they be incorporated into the RDF model theory; but I honestly do not 
believe that you will be able to come up with any such entailments. 
We have not managed to find any.
(Perhaps it might be worth emphasizing what 'entailment' means here. 
A entails B if B must be true whenever A is true. Entailment does not 
mean that one computational state leads to, or can be transformed 
into, another state. It is a process of accumulation of truths, not 
one of changes of state.)  And, to repeat, the RDF description of a 
bag tells you all you need to know (that it is a bag and what its 
members are) in order to process the information according to your 
understanding of what a bag is. It is true that RDF is not capable of 
formally *axiomatizing* the distinction between, say, bags and alts; 
but then to express such a distinction in an assertional language is 
known to be a fairly complicated business requiring a fairly complex 
language: and why would you *want* to axiomatize it in RDF, in any 
case?

>than rdf:Bag and rdf:Alt, if each time people used, e.g., rdf:Alt they
>would have to (in addition) state (somehow): "My interpretation is that
>rdf:Alt does not entail all its members but rdf:Bag does!"
>
>(*) This is the work and effort and definition: the work of coming up the
>concept rdf:Bag and saying that it is different from rdf:Alt. My opinion
>is that semantics should say exactly how they are or could be different,

The text is (RDF Semantics section 3.2.2, 
http://www.w3.org/TR/2003/WD-rdf-mt-20030123/#Containers):

"The intended mode of use is that things of type rdf:Bag are 
considered to be unordered but to allow duplicates; things of type 
rdf:Seq are considered to be ordered, and things of type rdf:Alt are 
considered to represent a collection of alternatives, possibly with a 
preference ordering. The ordering of items in an ordered container is 
intended to be indicated by the numerical ordering of the container 
membership properties."

Earlier, the text says:
"RDF provides vocabularies for describing three classes of 
containers. Containers have a type, and their members can be listed 
by using a fixed set of container membership properties. These 
properties are indexed by integers to provide a way to distinguish 
the members from each other, but these indices should not necessarily 
be thought of as defining an ordering of the container itself."

and:

"One should understand this RDF vocabulary as <em>describing</em> 
containers, rather than as a vocabulary for constructing them, as 
would typically be supplied by a programming language. On this view, 
the actual containers are entities in the semantic universe, and RDF 
graphs which use the vocabulary simply provide very basic information 
about these entities, enabling an RDF graph to characterize the 
container type and give partial information about the members of a 
container. "

---------------

Does this help?

Pat Hayes


-- 
---------------------------------------------------------------------
IHMC					(850)434 8903 or (650)494 3973   home
40 South Alcaniz St.			(850)202 4416   office
Pensacola              			(850)202 4440   fax
FL 32501           				(850)291 0667    cell
phayes@ai.uwf.edu	          http://www.coginst.uwf.edu/~phayes
s.pam@ai.uwf.edu   for spam

Received on Friday, 7 February 2003 12:24:52 UTC