- From: Karsten Tolle <tolle@dbis.informatik.uni-frankfurt.de>
- Date: Mon, 1 Sep 2003 15:13:17 +0200
- To: "Dan Brickley" <danbri@w3.org>, "Peter F. Patel-Schneider" <pfps@research.bell-labs.com>
- Cc: <bwm@hplb.hpl.hp.com>, <www-rdf-comments@w3.org>
Dear Dan, together with the current version of the RDF Primer the text of RDF Schema describes the current situation very well!-) My concern is more about the current situation itself. We have two structures, where the different container types have the benefit of serving a basic semantic but without the benefit of the lists (without a basic semantic) allowing to say all members are present. When a user now wants to create his RDF representation he needs to decide what to choose . if he knows about it . and each decision has its drawback. The best would be that in cases of well formed containers and collections we might be able to find a mapping between the two. But we would need an extension of the current structures for it. Possible solution (just a first try): a) Mapping from collection to container: Including the containers rdf:Bag, rdf:Seq and rdf:Alt to the class of rdf:List. This would enable the writer of RDF to construct an collection using the container types. In case all collection elements (the blank nodes) are of one container type (which can not be ensured), the structure could be transformed in an corresponding container. Open problem here would be how to handle the member numbers for sequences, e.g. a sequence just containing rdf:_2 and rdf:_5 is also valid. But I am sure that there are ways to solve this. b) Mapping from container to collections: A property could be introduced telling the counted number of members. If this information is present we can check if all members are around and could also construct the corresponding collection (as described in a). (I hope it is understandable. If not just let me know and I would try to generate some examples to explain it more in detail.) The goal at the end would be to have just one structure containing all benefits and to be able to transform existing RDF/XML structures into the new one. Greetings, Karsten P.S.: For the RDF/XML representation in the Editors Draft of RDF Schema I have the following comments: a) In the definition of rdfs:isDefinedBy the declaration of being of type Property is twice. b) For my opinion rdf:List should be also sub class of rdfs:Resource. c) The definition of rdfs:member is ok, but wouldn't it make sense to say that it is of type rdfs:ContainerMembershipProperty? > > Peter, Karsten, > > http://www.w3.org/2001/sw/RDFCore/TR/WD-rdf-schema-20030117/ (editor's > working copy) is now updated with our finalised Lists text. We propose > to close this LC issue on the basis of the new wording. Could you both > reply to this message to let us know whether the current text is > acceptable and resolves your issue. > > Thanks, > > Dan > > ps. Karsten, regarding your concern that we have both containers and > collections, those two mechanisms are described in more detail in the > Primer, see editor's copy section 4, > http://www.w3.org/2001/sw/RDFCore/TR/WD-rdf-primer-20030117/#othercapabiliti es > which provides a lengthier explanation of both constructs. > > > * Peter F. Patel-Schneider <pfps@research.bell-labs.com> [2003-06-23 13:29-0400] > > From: Dan Brickley <danbri@w3.org> > > Subject: Re: pfps-12 lists are not well formed > > Date: Fri, 20 Jun 2003 08:05:53 -0400 > > > > > * Peter F. Patel-Schneider <pfps@research.bell-labs.com> [2003-06-18 16:04-0400] > > > > This appears to be on the right track, but I have no way of viewing the > > > > other changes, and no way of viewing the changes in context, so I don't see > > > > how I can determine whether the changes are satisfactory. > > > > > > Fair point. The WD text has changed little since last call. I have some > > > outstanding edits derrived from other LC comments, but nothing (to my > > > knowledge) which should impact on this issue. > > > > > > http://www.w3.org/2001/sw/RDFCore/TR/WD-rdf-schema-20030117/ is the > > > LC text plus some minor edits: > > > > > > ---------------------------- > > > revision 1.12 > > > date: 2003/04/11 13:58:50; author: danbri; state: Exp; lines: +1 -1 > > > Fixed typo per: > > > http://lists.w3.org/Archives/Public/www-rdf-comments/2003JanMar/0416.html:wq > > > ---------------------------- > > > revision 1.11 > > > date: 2003/01/24 18:03:53; author: danbri; state: Exp; lines: +51 -43 > > > fixed date errors. > > > ---------------------------- > > > > > > I plan to make some more modest updates in-place at that URL (with > > > 'status' updated accordingly to avoid confusion), until we agree new URIs for > > > the next batch of TR-published RDFCore docs. > > > > > > My editorial todo list is based on the issue closures linked from > > > our LC issue list, see http://www.w3.org/2001/sw/RDFCore/20030123-issues/#schema > > > > > > I hope this clarifies the state of the doc for you. If I commit changes > > > to the spec I'll make these explicit so you can see what's what. > > > > > > cheers, > > > > > > Dan > > > > Good. I await the changes to the appropriate sections of the document for > > final approval then. > > > > Peter F. Patel-Schneider > > Bell Labs Research > > Lucent Technologies > > >
Received on Monday, 1 September 2003 09:13:58 UTC