On Saturday, December 18, 2004, 10:44:29 AM, Danny Ayers wrote:
> I've not really spent any time around provenance, but three approaches > spring to mind > ... Perhaps part of the problem is related to modelling dynamic data using a static model. Another approach would be to model those updates, rather than attempt to resolve them into a static model at the RDF level. This would require a vocabulary that explicitly supports an indirection between the item in the feed, and a particular state of the item in the feed. Eg, if RSS 1.0, looked something like this: <rss:item rdf:about="http://example.com/myitem";> <ex:itemState> <ex:ItemState> <rss:title>whatever</rss:title> <!-- more properties --> <ex:modified>2004-12-18</ex:modified> <ex:source rdf:resource="http://example.com/source"; /> </ex:ItemState> <ex:itemState> </rss:item> Then it would be possible to attach provenance information and temporal information to a particular version of an item in the feed. The choice of which state of an item to display could then be done at the application level, and it would be fairly easy to support extensions like delayed publishing, expiry, and deletion of items. -- Dave
