- From: Robert Sanderson <azaroth42@gmail.com>
- Date: Wed, 8 Aug 2012 16:01:59 -0600
- To: Stian Soiland-Reyes <soiland-reyes@cs.manchester.ac.uk>
- Cc: Bob Morris <morris.bob@gmail.com>, public-openannotation <public-openannotation@w3.org>
On Wed, Aug 8, 2012 at 9:26 AM, Stian Soiland-Reyes <soiland-reyes@cs.manchester.ac.uk> wrote: > On Tue, Aug 7, 2012 at 8:33 PM, Robert Sanderson <azaroth42@gmail.com> wrote: >> I'm not sure *how* it would be modeled though. >> >> A suggestion was raised to get rid of equivalent and recommend either >> skos:exactMatch or skos:closeMatch. How do people feel about these >> three options? > +1 to avoid yet another equivalence notion! > However, what do we need it for? For mirroring annotations? For three primary scenarios: 1. An annotation is created by a client that cannot assign HTTP URIs by itself, and hence the annotation's identifier is a UUID. The server then creates an HTTP URI for the annotation and asserts the equivalence (in some way) between the UUID and the new HTTP URI. 2. An annotation is created by a client that embeds some of the resources using the Content in RDF specification. It is published to a server which then extracts the resources and gives them their own HTTP URIs allowing them to be dereferenced separately. The server then asserts the equivalence between the UUID of the Content in RDF resource and the new HTTP URI. 3. A second harvesting/subscribing server collects annotations from the first server in 1 and 2. It then republishes the annotations at URIs in its own domain and asserts equivalence between the URI on the original server and the URI on the new server. And all of this is in order to prevent another service from harvesting all of the above resources and ending up with multiple copies of them to store and index. > I would question whether such annotations are equivalent or even close > match - because it could have later change upstream or even where it > is now. I would be happy with close. If they divert to the point where they're no longer close, then they probably should be treated separately. > It is more like it has been derived? I would use prov:wasQuotedFrom > from PROV-O (it's like a full quote), combined with prov:alternateOf > to show that they were somewhat interchangeable at the time. prov:alternateOf seems like a better relationship than oa:equivalent or skos:closeMatch. Rob
Received on Wednesday, 8 August 2012 22:02:28 UTC