- From: Dan Brickley <danbri@w3.org>
- Date: Mon, 8 Nov 2004 16:51:09 -0500
- To: Phil Archer <phil.archer@icra.org>
- Cc: Quatro Public list <public-quatro@w3.org>
* Phil Archer <phil.archer@icra.org> [2004-11-08 21:09-0000] > > As promised I've put a few use cases online. I'll try and think of some > others but these are the top of my list. > > I've written one document that describes the use cases and gives some > detail of what is expected, then created actual test data for each one. > > Please see http://www.icra.org/projects/quatro/techdiscussion/usecases/ > > There must be a reason why the RDF instance at > http://www.icra.org/projects/quatro/techdiscussion/usecases/exemplar/labels.rdf > doesn't validate but I can't see it... Hi Phil, I've had a quick play with the example, tweaking it to fit RDF/XML's so-called 'striped' syntax. Basically this means that XML elements and attributes altenatively stand for nodes and edges in the underlying graph. I've also changed the case conventions to use capital letters for classes (ie. categories, resource types), and lowerCase notation for relationships/properties. I have marked up the application rule section using RDF's list construct, ie the parseType="Collection" attribute. I have not added any ordering around the application rule and match construct, though I think we discussed doing so. Revised file, plus the graphic from the RDF validator, is at http://www.w3.org/2004/03/quatro/tests/week1/ Backgrounder on RDF 'striping' that I wrote a couple years back, might be useful, is at http://www.w3.org/2001/10/stripes/ I've not had a chance to think thru details of http://www.icra.org/projects/quatro/techdiscussion/usecases/ yet, but I think the local override case will be the one that exercises us the most. It's good to write up such use cases even if we end up not satisfying them all, of course. A technical term floating around nearby to this issue is "monotonic" (see W3C glossary, http://www.w3.org/2003/glossary/keyword/All/?keywords=monotonic ). We could exchange descriptions of non-monotonic rules using RDF, but RDF's design itself is monotonic: it guarantees that we if we believe two documents, then combining them shouldn't produce something we disagree with. There are loopholes (eg. exchanging descriptions of descriptions) but in general I think there are things in favour of not having a defaults mechanism. cheers, Dan
Received on Monday, 8 November 2004 21:51:10 UTC