Re: New Intro to RDF

As just pointed out - this is the "URI Crisis" perma-thread. In direct
response, the "use a new URI scheme" versus "use http:" issue is quite
interesting, and myself and Henry Thompson propose in "Web Proper Names"
a rather common-sense way using http URIs with a special type of RDDL
file to solve the URI crisis issue, as well as giving a background on
various theories of naming and reference. I do think there is a definite
advantage in using URIs that allow one to "get more information" about
what someone means by a piece of data, even if one cannot directly
retrieve the data itself.  For more, see:
http://www.webpropernames.org/paper/

Using a differing URI Scheme (such as tag) merely sidesteps the  "URI
Crisis." This is just a specialisation of what is known asthe
"Symbol-Grounding Problem" in philosophy - which is how does an
arbitrary symbol get its meaning? In computer programs one can make the
"behavioral" case - a symbol gets its meaning from what it *does*, as is
does in operational semantics ala XQuery. In KR systems, one generally
has to appeal to "what the author meant" and the fact that the what the
author meant (ala TimBL) and the inferences allowed by the KR system
line up. On the Web in general, it is very hard to say, since unlike
traditional KR and computer programming you have this "information"
(pictures, text, more RDF even!) that a URI can possibly retrieve for
you - and with things like pictures and text, while they have no
behavior, they carry lots of "social meaning."

 I think one advantage of using http URIs is to make this work *for* you
rather than against you, by making sure the text and pictures retrieved
by your URI by either explicitly controlling what information the URI
delivers or by having a good chance (such as w3.org being about "W3C")
of conveying what you mean. My main issue with tag URIs is that I see no
reason why one wouldn't just want to use http URIs in case *later* you
wanted a Web representation of what you were talking about to be retrieved.

                              -harry

Sandro Hawke wrote:

>>Charles McCathieNevile wrote:
>>    
>>
>>>(tag: and urn: URIs fail to impress me as being useful until there is an 
>>>HTTP service that can resolve them
>>>      
>>>
>>The whole point of tag: and urn: is to have a system of creating 
>>identifiers for things that have no representation on the web.
>>    
>>
>
>Not at all.  There are several ways to use http URIs as identifiers
>for things which have no representation, and they're very useful.
>They let you "follow your nose" and easily find information about
>something just by knowing its identifier.
>
>The point of tag: URIs is to allow you to identify things when you
>don't want to allow people to "follow their nose", because you are not
>willing to maintain a web presence.
>
>       -- sandro
>
>
>  
>

Received on Monday, 10 October 2005 22:24:17 UTC