Re: What if an URI also is a URL

Hi,

One more solution (at Web architecture level):

1/ Extend the http protocol to cover the reference nature of http uri.

Simply add one method "what".

Existing http methods,such as get/put/post ..., have the nature of "access". In contrast, the proposed "what" method returns RDF description of the requested http uri, with the nature of uri reference.

Intuitively, the "what" request method is used to ask "what the requested http uri is or refers to"

//Similar to mget method in URIQA[2], but without other methods and CBD requirement.

Any Semantic Web server should support the "what" method.

2/ Web browsers are enhanced with the capability to issue "what" requests (on behalf of users) and/or extend XHTML to express the reference nature of uri.

In addition to the ordinaire address control with the "access" nature, any Semantic Web browser should support "referential address" control, which will tell users what the requested uri is or refers to (RDF data or transformed to XHTML via XSL).

A possible syntactic sugar for extending XHTML, <r href="uri"> text </r> .
Whenever a user clicks the "referential link", the browser issues a "what" request, and then the authority (host) returns as a response an RDF description of the requested uri.

//end of the proposal

Why such a solution?

1/ With the Semantic Web, http uri have two different natures: Reference and Access.

//See Pat Hayes [1]. 

2/ Users or machines need to know what a http uri denotes or refers to, with great confidence.

Otherwise, users or machines can't get a trustful description of a uri, and the reuse of URI on the Semantic Web (and the Web of Data) would be VERY difficult to achieve (if not impossible).

Think about why the Web pages can be linked naturally and easily.   

3/ A case study: Tim's identifier (URI) used in the Semantic Web (SW).

http://www.w3.org/People/Berners-Lee/card#i


http://projects.semwebcentral.org/users/timbl


http://inferenceweb.stanford.edu/registry/PER/BERNERS-LEE.owl#BERNERS-LEE


http://ontoworld.org/wiki/Special:URIResolver/Tim_Berners-2DLee


http://norman.walsh.name/knows/who#tim-berners-lee


//The list is NOT complete

Some triples on the Semantic Web.

<ns1:User rdf:about="http://projects.semwebcentral.org/users/timbl">
  <ns1:name>Tim Berners-Lee</ns1:name> 
</ns1:User>

<iw:Person rdf:about="http://inferenceweb.stanford.edu/registry/PER/BERNERS-LEE.owl#BERNERS-LEE" >
  <iw:name>Tim Berners-Lee</iw:name> 
 </iw:Person>

So many URIs for one thing on the SW!!! (even for Sir Tim Berners-Lee)

This situation becomes a big impediment to the developement of SW (as a Web of Data). 

One may argue that matching URI and using owl:equivalentTo (or sameAs) could alleviate this situation. Besides, searching URIs on the SW (such as SURI described in our submission to ISWC2007) would promote the reuse of URI on the SW. 

However,I think, this issue must be considered at architecture level, and should be investigated in the social context(trust,confidence). And It's an urgency!

Any comments is welcome!  



Yuzhong Qu
http://cse.seu.edu.cn/people/yzqu/

http://iws.seu.edu.cn


Related work:
[1]Pat Hayes, In Defense of Ambiguity
 http://www.ibiblio.org/hhalpin/irw2006/presentations/HayesSlides.pdf

[2]URIQA 
http://sw.nokia.com/uriqa/URIQA.html



----- Original Message ----- 
From: "Sandro Hawke" <sandro@w3.org>
To: "r.j.koppes" <rikkert@rikkertkoppes.com>
Cc: "Tim Berners-Lee" <timbl@w3.org>; <semantic-web@w3.org>; <swick@w3.org>
Sent: Thursday, June 07, 2007 3:29 AM
Subject: Re: What if an URI also is a URL


> 
> 
> "r.j.koppes" <rikkert@rikkertkoppes.com> writes:
>> Then please read "look up" instead of "access", I am talking about 
>> visiting http://www.example.com/mophor#me as a web page, following a 
>> link with this as a reference or putting it in an address bar of a browser.
>> 
>> This action only identifies http://www.example.com/mophor as it is this 
>> URL that makes the server return a 200.
>> 
>> The URI http://www.example.com/mophor#me is not identified as a resource 
>> at this moment.
>> 
>> But what if I DO identify it, by means of an RDF triple stating that 
>> this URI (http://www.example.com/mophor#me) defines me (by linking it to 
>> my social security number or whatever)
>> 
>> As of what I understand, up till now there are no problems. 
>> http://www.example.com/mophor is identified as being a web page, 
>> http://www.example.com/mophor#me is identified as being me.
>> 
>> But if, on the web page http://www.example.com/mophor there is a section 
>> with id "me", how do I refer to that particular section in the web page 
>> in a RDF document (which might contain anything, even unrelated to me as 
>> a person)? How do I make sure that the reader (machine / human) 
>> interprets this reference as being a web location (fragment in web page) 
>> instead of the thing, me.
> 
> This is a problem.  You've described it quite nicely.   Various
> solutions have been proposed over the years.   I think the two leading
> contenders are:
> 
>    1.  Don't Do That.   Don't have that "#me" anchor on that page.
> 
>    2.  Use a "303 See Other" redirect, based on content negotiation, so
>        that when someone asks for "http://www.example.com/mophor" they
>        get redirected to one of two other pages, based on the content
>        type they ask for:
>           text/html -> "http://www.example.com/mophor-text"
>           application/rdf+xml -> "http://www.example.com/mophor-data"
>        The RDF page would talk about 
>    "http://www.example.com/mophor#me" as you.
>        The HTML page would have a "me" fragment, but the base URI of
>        the page with that fragment would be interpreted AFTER the
>        redirect, so it would have the URI:
>           http://www.example.com/mophor-text#me

> 
>        (At least I think that's how it would work.  I haven't dug into
>        this one in a while.)
> 
> For most people, Option 1 seems to suffice.  The motivation for Option
> 2, for me, is that I think all RDF URIs should work in browsers.  That
> is, I should be able to take any RDF URI and paste it any reasonable
> browser, and get some good HTML about the identified thing.
> 
> As far as I know, no one who is hosting an RDF vocabulary actually does
> this yet, alas.
> 
> Now that I think about it, this should probably be added to the Recipes
> for Publishing Vocabularies....  [1]
> 
>     -- Sandro
> 
> [1] http://www.w3.org/TR/swbp-vocab-pub/ 
> 
> 
>

Received on Thursday, 7 June 2007 06:02:37 UTC