- From: Nandana Mihindukulasooriya <nmihindu@fi.upm.es>
- Date: Thu, 17 Jul 2014 20:54:43 +0200
- To: public-lod <public-lod@w3.org>
- Message-ID: <CAAOEr1=1tYT0ZiuA+C8Hqs6S4ubYqRhJEmqXO9Xgyn0Cx9Y4Qg@mail.gmail.com>
Hi Pieter, On Thu, Jul 17, 2014 at 7:36 PM, Pieter Colpaert <pieter.colpaert@ugent.be> wrote: > Hi Nandana, > > Thank you a lot for your clear reply! > > > On 2014-07-17 19:17, Nandana Mihindukulasooriya wrote: > > Hi Pieter, > > If we still stick with URIs (as a name but not a locator) [1] but with a > different scheme, say "things" or something, your solution will still work > the same, right? There are already URN/DOI to URL resolvers [3], so > similarly but rather than using a service, your URIs identifying real world > things will use a convention to resolve them to information resources by > converting, say things:{foobar} to http://{foobar}, when one have to do a > lookup. > > > Correct! "thing:" could be the protocol of the real world: thing:A can > shake hands with thing:B, http://A can serve the fact that thing:A shook > hands with thing:B over HTTP. I like it! > Apparently this has been considered and discarded with the argument of building the Semantic Web on top of what was already available back then. http://www.w3.org/DesignIssues/HTTP-URI.html#L920 Along with the URI collision that David Booth pointed out, we have the Indirect Identification use case (i.e., the context defines what the URI identifies). Probably this works well for humans but not so well for machines. http://www.w3.org/TR/webarch/#indirect-identification Best Regards, Nandana
Received on Thursday, 17 July 2014 18:55:27 UTC