- From: Bernhard Schandl <bernhard.schandl@univie.ac.at>
- Date: Mon, 8 Mar 2010 14:54:02 +0100
- To: Hugh Glaser <hg@ecs.soton.ac.uk>
- Cc: Kingsley Idehen <kidehen@openlinksw.com>, Aldo Bucchi <aldo.bucchi@gmail.com>, Linked Data community <public-lod@w3.org>
Hugh, > Maybe I am still misunderstanding, but I think that you rare still saying > that a urn/doi approach is compatible with Linked Data, or at least is not > harmful. > I think differently. > urn/doi is harmful - once it comes into existence, it is very hard to avoid > the problems of people using them, and then you have to start working out > where the server might be. > I am quite happy with people passing round http://foo.com/bar/urn:baz.quex, > as this is resolvable; I just don't want things that don't use http. > > I think Bernhard's questions suggest that your comments might have been > misinterpreted by him as I did, that urn: is acceptable as a Linked Data > URI. I didn't understand it that way. My question came from my objections against storing full dereferenceable HTTP URIs in a database, as this makes it difficult to migrate and distribute data. My question was: are there any objections against using URNs internally (i.e., within a database/triple store/file system/whatsoever), which are dynamically rewritten to HTTP URIs when they are served to the outside? (btw, this is exactly what e.g., Pubby [1] does.) So it's a question of internal vs. external identifiers -- both have their (different) purposes. And second question: if you already have the URNs, are there objections against publishing them *alongside* the HTTP URIs (using owl:sameAs), and therefore allowing them to be used as strong indicators for equivalence when linking datasets? Think of ISBN numbers published as URNs [2] alongside with resolvable HTTP URIs. There is a known schema for ISBN URNs, so why not reuse it? [1] <http://www4.wiwiss.fu-berlin.de/pubby/> [2] <http://www.apps.ietf.org/rfc/rfc3187.html> Best Bernhard
Received on Monday, 8 March 2010 13:54:35 UTC