- From: carmen <_@whats-your.name>
- Date: Wed, 18 Jul 2007 16:02:46 -0400
- To: semantic-web@w3.org
On Wed Jul 18, 2007 at 05:25:00PM +0200, Jérôme Mainka wrote: > > Hi, > > I try to specify some kind of templating in RDF. My goal is to compose the > value of a property from other properties. Here is an example : this sounds more like 'computed values' than templating. you might want to check out cells (Lisp), google for cells-gtk. you can surely do some fantastic things compiling that with Wilbur. Wilbur is also fairly advanced in its own right, and if i recall is geared more towards these sort of things than most RDF frameworks which tend to just offer classes for the basic structures > My first attempt was to write a SPARQL rules like this : i'd definitely check out something besides sparql. something where the triples are efficiently exposed in a language and you have easy feedback into the model. (ie, no 'stored procedures' in sparql, rather the indexing/querying substructures of the sparql engine are available to you to play with) there are some pdfs on stored procedures in sparql, and also a DSL geared towards rdf models: http://ripple.projects.semwebcentral.org/ > But this kind of construction doesn't work with a partial knowledge. In my > example, if :prop2 is missing, the premises aren't satisfied. there is OPTIONAL in SPARQL. i cou ldnt get OPTIONAL working with CONSTRUCT in redland, or find much specifics on whats supposed to happen in that case in the spec on w3.org.. > > Am I pushing RDF too far? no > Should I modify my model? maybe > > Thank you for your suggestions. thanks for your curiosity, report back your progress on computed properties, if you dont mind! > > Jérôme Mainka > > -- > Antidot - solutions de recherche d'information > gsm: +33 6 62 79 09 74 > ip : +33 9 50 28 37 46 > fax: +33 9 55 28 37 46 >
Received on Wednesday, 18 July 2007 20:02:54 UTC