- From: Pierre-Antoine CHAMPIN <pachampi@caramail.com>
- Date: Fri Apr 21 08:39:49 2000
- To: Jonas Liljegren <jonas@paranormal.se>, RDF Intrest Group <www-rdf-interest@w3.org>
- Message-ID: <956318773019830@caramail.com>
> > technically, this won't work > That is a matter of interpretation. My Schema editor works in that > way. But I guess that many RDF parsers doesen't do it. There must be > many sorts of inference rules in RDF. well yes, it is a matter of interpretation, so RDFS should not be ambiguous on the subject ! > One way to make all this explicit would be to create a schema. > <awful schema> ;) > ... But I would prefere to magicaly let the range always include the Container class. sure ! that sounds like a sane solution. > That could maby be done explicitly by constructing a new > ConstraintProperty containerRange. So if you wanted the range of > myprop to always be a sequence of literals, you could say: <rdf:Property> <rdfs:range rdf:resource="&rdf;Literal"/> <rdfs:containerRange rdf:resource="&rdf;Seq"/> </rdf:Property> I like this one too... But would the absence of ocntainerRange property mean : - any type of container is allowed (consistent with rdfs:range semantic) - no type container is allowed (unconsistent, but how express that another way ?) ? interesting issue, anyway, since it allows - to explicitly declare properties admitting containers as their value - to force type of container's items with the rdf:range mechanism this would be quite equivalent to OO techniques' mulivalued attributes. Pierre-Antoine _______________________________________________________ Vendez tout... aux enchères - http://www.caraplazza.com
Received on Friday, 21 April 2000 08:39:49 UTC