Re: What if an URI also is a URL

In my opinion, this thread shows that there is still an unsolved problem of 
fundamental importance to the semantic web. The problem is that there is no 
official or well accepted de facto method for establishing and conveying the 
reference of a URI that is not a static web page. This is not a problem with 
just the semantic web, it is a problem for all beings that use languages. 
But it is also a problem faced by the semantic web initiative. Nor is this a 
"philosophical rat hole", without relevance to real world problems. Instead 
it is a daily struggle for all beings that use language. Communication with 
symbolic languages is hard, and it is only the unfathomably powerful 
linguistic abilities of humans that makes it possible at all. My post 
http://kashori.com/2006/06/anatomy-of-reference.html discusses some of the 
issues of reference.

There are reasons for this problem:
1. Symbols and conceptualizations are discrete but the world they are about 
is continuous. The lines we draw around the things or concepts we want to 
refer to are thoroughly human, arbitrary, and ad hoc. I give examples at 
http://kashori.com/2006/07/ambiguity-and-identity.html
2. The interpreter of the reference of a URI is entirely independent of the 
publisher of a reference. We can't control how clients will interpret a 
symbol or a URI. They do want they want. And they can only work with what 
they've got.
3. Every agent in a conversation has its own view point. There is no 
external, absolute viewpoint with the one true apprehension of the true 
nature of a thing or a class of things.
4. Every symbol or URI refers only in the context of, or background of, an 
entire semantic web of experiences of the world. See John Searles examples 
http://kashori.com/2007/04/john-searles-hypothesis-of-background.html. I 
express this further in 
http://kashori.com/2006/07/words-or-uri-as-locations-in-fabric-of.html.

John Black

Tim Berners-Lee wrote
>
> On 2007-06 -09, at 06:39, r.j.koppes wrote:
>
>> The thing that still puzzles me is the following: a URI is unique  in the
>> thing it represents, indeed, it is a resource identifier. a URL is  only
>> unique in the location and can mean very different things
>
> No no no no no.
> The term "URL" is NOT patt of the architecture. It has no well- defined 
> meaning.
> (It has been used to mean "URIs which use FTP or HTTP schemes as  opposed 
> to URN schemes"
> or "URIs which might break" or URIs of documents") It would be  simplest 
> if you stop using it
> and only use the term URI.  The use of a string as a 'URL' is not 
> different from its use as a URI.
>
>
>> for example the URL http://www.example.com can be a html or xhtml
>> document, depending on the client accessing it.
>
> Yes. Please see http://www.w3.org/DesignIssues/Generic.html for a  full 
> discussion  of generic URIs.
>
>> Furthermore, the string
>> (this might be crucial) http://www.example.com might act as openId url
>> for someone.
>
> No. It cannot identify both a document and a person.
>
> Tim Berners-Lee
>

Received on Monday, 11 June 2007 02:17:54 UTC