Re: Properties of Resources that Need More than One URI to be Identified

I'm going to ignore the vagaries of the file: scheme, and try and respond 
more directly.

1.  If the only way you know of identifying a resource is by the 
conjunction of two related values, then RDF has perfectly respectable way 
to do this:

(Using Notation3):

   _:f a :File ;
       :localName <file:///path/name> ;
       :hostLocation :some_URI_used_to_represent_a_host_location .

This is the sort of approach that I understand underlies Guha's "reference 
by description" work [1].

2.  The fact that the only way *you* may have of identifying a resource is 
by using two values doesn't mean that it's not possible to mint a new URI 
for every resource thus described.  Here's how I do it:  given any two 
URIs, u1 and u2, I construct a new URI:

     http://id.ninebynbine.org/uripair?u1=uuu1?u2=uuu2

where 'uuu1' and 'uuu2' are the uris u1 and u2 with some escaping applied 
as needed to make the whole URI valid and unique with respect to u1 and u2.

I don't claim this is useful, just possible.

#g
--

[1] http://tap.stanford.edu/tap/rbd.html


At 17:57 10/11/03 -0500, James Cerra wrote:

>One basic premise of RDF is that any non-anonymous resource can be 
>identified by a URI.  But what if that wasn't true?  What if there was a 
>resource that needed _two URIs_ for it to be identified?  In this case, 
>each URI, in the context of itself, could describe something completely 
>different from the thing identified by both URIs!  What would be the 
>properties of such a resource?
>
>For instance, say I wish to identify a file on my computer on another 
>(internet and network disabled) computer.  Note that I don't care about 
>getting the representation for that file, just identifing it.  I can't 
>just use the file: protocol since that is useful only in the context of a 
>specific machine.  However, I could use the file: scheme if I also used an 
>address: scheme (that latter specifies the location on the Earth of the 
>computer) in context.  Order is important, the address: URI wouldn't make 
>sense in the context of the file: uri.
>
>Unfortunately, RDF doesn't seem to handle reosurces that need multiple 
>URIs for identification.  This is troubling to me, since I have come 
>across other situations where I need more than one (or two!) URIs for a 
>concrete identification.  It would be nice; however, if there was a way to 
>combine several URIs into one identification string.
>
>Any suggestions or help?
>
>-- Jimmy Cerra
>
>_________________________________________________________________
>MSN Shopping upgraded for the holidays!  Snappier product search... 
>http://shopping.msn.com

------------
Graham Klyne
For email:
http://www.ninebynine.org/#Contact

Received on Tuesday, 11 November 2003 08:04:50 UTC