Re: Clarifications needed for the Collection construct

>Dear Pat,
>
>sorry for the late reply.
>
>I understand why this new collection is included and I agree that there
>was the need to extend RDF to allow to integrate an upper bound for
>entailing. What I do not like in the found solution is that we now have
>the containers with at least a minimum of semantic for it and we have
>the collection in addition.

I don't exactly LIKE it myself :-)

>The collection comes with no semantic and
>there are no rules given when to use a collection or a container. This will
>irritate the people that want to create some RDF and will also makes it
>more difficult to deal with it. This is contra productive!
>
>There could have been a property introduced (I think it would still 
>make sense)
>that gives the length (number of members) for a container. Of course the
>processing would be different but it would also guarantee an upper bounding.

We did discuss this some time back. The problem was, for some people, 
that having a special 'last thing' in a container would mean that 
this thing couldn't itself be a real member of a container.

Another possibility was to associate a number (of items) with a 
container, but that requires incorporating a numerical datatype into 
core RDF.

Still another , which we didn't think of at the time, would be to 
have a property of membership properties, so that one could write say

_:x rdf:lastMemberProperty rdf:_17 .

to indicate that _:x has only 17 things in it. But this kind of 
'higher-level' assertion is likely to be confusing, and to break a 
lot of deployed UML-style code.

None of these is completely satisfactory; and there were no users 
clamoring to have this feature added in any case, so we felt it was 
better, and closer to our charter, to adopt the strategy that if it 
isn't broken, to not try to fix it.

>
>This way the collection would not be needed any more.


The real point is that the collection syntax was explicitly requested 
by another WG, so we have to include that, whatever we do about 
containers. And given that the collection syntax is available, it 
doesn't seem worth 'fixing' the container syntax; particularly as our 
WG charter explicitly asks us not to make any changes that are not 
absolutely necessary; and particularly as we had no existing users of 
the container syntax who positively requested that it be changed.

>In case the way of
>structuring the collection has additional advantages we could think of a
>mapping between containers and collections. Š we would need a container
>without any meaning and attributes for collections to represent the semantic
>for the existing container types.

I tend to agree that this would be a more perfect world, but the 
exigencies of history have landed us in this one, and it doesn't seem 
to be bad enough that it is worth the considerable effort to change 
it.

Our process requires one of three possibilities at this point. Your 
reply did not say whether or not you 'accept' our decision. Finding 
it acceptable is not necessarily approving of it wholeheartedly, 
notice. If you do accept our decision, then we can close this issue. 
If you do not, then either we re-open the discussion, or we formally 
note, and archive, that you do not find our decision acceptable.

I think it is very unlikely that we will re-open this issue at this 
stage.  Your call.

Please reply to this email, copying www-rdf-comments@w3.org 
<mailto:www-rdf-comments@w3.org>  indicating whether this decision is 
acceptable.

Pat


>
>Cheers,
>Karsten
>
>___________________________________
>Karsten Tolle
>----- Original Message -----
>From: "pat hayes" < phayes@ai.uwf.edu <mailto:phayes@ai.uwf.edu> >
>To: < tolle@dbis.informatik.uni-frankfurt.de 
><mailto:tolle@dbis.informatik.uni-frankfurt.de> >
>Cc: < www-rdf-comments@w3.org <mailto:www-rdf-comments@w3.org> >
>Sent: Tuesday, April 22, 2003 12:30 AM
>Subject: Re: Clarifications needed for the Collection construct
>
>>  Dear Karsten'
>>
>>  re. your comment at
>>
>>  <http://lists.w3.org/Archives/Public/www-rdf-comments/2003JanMar/0328.html>
>>
>>  The general point you make here can, I think, be summed up by saying
>>  that RDF does not impose any well-formedness conditions on its
>>  collection vocabulary, so that it is possible to write RDF graphs
>>  which make no 'sense' relative to the indicated intended
>>  interpretation of the collection vocabulary. This is correct, as RDF
>>  provides no syntactic constraints of this kind.
>>
>>  You also ask about the reason for introducing the collection
>>  vocabulary. The collection vocabulary was requested by the DAML joint
>>  committee and the Webont WG. The difference between the collection
>>  and container vocabularies lies in the fact that it is possible to
>>  write an RDF graph which entails that the number of things in a
>>  collection has an upper bound, while it is not possible to do that
>>  with the container vocabulary.
>>
>>  We have not made any changes to the document as a result of your
>  > comment. Please reply to this email, copying 
>www-rdf-comments@w3.org <mailto:www-rdf-comments@w3.org>
>  > indicating whether this decision is acceptable.
>>
>>  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 <mailto:phayes@ai.uwf.edu> 
>><http://www.coginst.uwf.edu/~phayes>
>>  s.pam@ai.uwf.edu <mailto:s.pam@ai.uwf.edu> for spam
>>
>>


-- 
---------------------------------------------------------------------
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, 2 May 2003 19:19:42 UTC