- From: Gabe Beged-Dov <begeddov@jfinity.com>
- Date: Sat, 04 Dec 1999 10:02:14 -0800
- To: Sergey Melnik <melnik@DB.Stanford.EDU>
- CC: RDF Interest Group <www-rdf-interest@w3.org>
Sergey, Thanks for restarting the API discussions. Let's shoot for end of 1999 rather than the end of the millenium :-). I like the API and look forward to seeing the repackaged sample implementation. I have two areas of comments below . Please excuse any sloppyness in description since I want to get something out before I'm overtaken by weekend commitments (leaf raking :-). Models are Resources ================== Many of the participants have indicated their desire for being able to treat models as resources. I just wanted to point out what may be obvious which is that each set operation returns a new model. Each of these result models should also have their own URI. This falls under the issue of NoName resources. Triples as Resources ================= I think that the URI for triples should be totally separate from the XML serialization. Jonas describes an approach using XPointer that is heavily dependant on a particular XML serialization of a model. I think the approach that is serialization independant is more robust. The URIref should be composed of the URI of the model with a UUID fragment identifier for the triple using something like one of Sergeys' approaches. Once you pick a URIref for the triples you can serialize them to XML using the basic syntax which allows a property to have an rdf:ID. This assumes that the triple has the same URI as the model that is being serialized and that this URI will continue to serve as the base URI for the model for future deserialization. This property ID seems to be lost by the current crop of RDF parsers but that could be an enhancement request once triples are resources. There is the question of whether triples should be traceable independant of the model that they are contained in. I.e. if you union two models should the triples in the resulting model have URIrefs whose URI is that of the origin models. If this is the case, you could no longer use the current RDF MS property ID to losslessly transport the result model. Cordially from Corvallis, Gabe Beged-Dov http://www.jfinity.com/gabe
Received on Saturday, 4 December 1999 14:11:17 UTC