Your citation supports my point. The question of "dereferenceability" only arises in the context of Linked Data.
See the RDF recommendations:
"Perhaps the most important characteristic of IRIs in web architecture is that they can be dereferenced, and hence serve as starting points for interactions with a remote server. This specification is not concerned with such interactions. It does not define an interaction model. It only treats IRIs as globally unique identifiers in a graph data model that describes resources. However, those interactions are critical to the concept of Linked Data [LINKED-DATA], which makes use of the RDF data model and serialization formats."
The University of Virginia Library
On Jul 17, 2014, at 10:39 AM, Karen Coyle <[log in to unmask]> wrote:
> On 7/17/14, 7:07 AM, [log in to unmask] wrote:
>>> URNs are by definition URIs, but in the semantic web context only http URIs are used for subjects and predicates (although, beyond the "use http URIs" statement by TBL, I don't see an absolute restriction in the RDF documentation).
>> You won't find any absolute restriction because there isn't one.
> Actually, there is the question of "dereferenceability" -- and that at the moment, on the web, only http uris are dereferenceable .
> I may have spent too much time on Wikipedia lately, but I think it's a good practice to provide a citation or two to support one's statements. Otherwise, we're just slinging unsupported opinions.
>> This is a key distinction between RDF (which exists independently of the Web) and "Linked Data" as we usually speak of it (which does not). It's perfectly legitimate to produce RDF using non-HTTP URIs and people have been doing it for many years. It's also perfectly legitimate to publish such RDF on the Web, and people have been doing that, too. But it's not Linked Data in the sense in which TBL used the phrase, or at least, not very good Linked Data. It's when we turn from the graph (RDF) to the distributed graph (Linked Data) that using HTTP URIs becomes so interesting, because HTTP is highly scalable and allows us to distribute the graph very widely and robustly.
>> A. Soroka
>> The University of Virginia Library
>> On Jul 17, 2014, at 10:00 AM, Karen Coyle <[log in to unmask]> wrote:
>>> On 7/17/14, 3:43 AM, Thomas Berger wrote:
>>>> Sure. But Rob's (absolutely valid) point are the inconsistencies arising
>>>> from resource URIs we want to distinguish as "identifiers" simultaneously
>>>> used as resource URI for the bf:Identifier as such. Making statements about
>>>> some URIs (as URIs as in contrast to the resources they represent) would
>>>> constitute a more subtle form of the same(?) fallacy and is likewise absolutely
>>>> not admissible.
>>>> My point of view however would be to simply dispense n URIs into m bf:Identifier
>>>> containers and keep them striclty in object position there...
>>> Which *you* can do, but you cannot prevent anyone else from using them "in the subject position." Therefore, in the open web, you cannot enforce this consistency.
>>> I note that all of your examples use URN forms, not http URIs, which are the LOD standard. That could mean that we are not talking about the same thing. URNs are by definition URIs, but in the semantic web context only http URIs are used for subjects and predicates (although, beyond the "use http URIs" statement by TBL, I don't see an absolute restriction in the RDF documentation).
>>> What I believe you are proposing is the same that I proposed in the schema.org variant , which is to have an "identifier" property for those identifiers that CANNOT be used as subjects in RDF statements. If that is the case, then it is essential that no URIs are used as objects of that predicate.
>>>  http://www.w3.org/community/schemabibex/wiki/Identifier-2
>>> Karen Coyle
>>> [log in to unmask] http://kcoyle.net
>>> m: 1-510-435-8234
>>> skype: kcoylenet
> Karen Coyle
> [log in to unmask] http://kcoyle.net
> m: 1-510-435-8234
> skype: kcoylenet