W3C home > Mailing lists > Public > www-rdf-interest@w3.org > April 2005

Re: SemWeb Non-Starter -- Distributed URI Discovery

From: Charles McCathieNevile <charles@sidar.org>
Date: Sun, 03 Apr 2005 14:20:25 +1000
To: "Josh Sled" <jsled@asynchronous.org>, "Rodrigo Dias Arruda Senra" <rsenra@acm.org>
Cc: dviner@apache.org, www-rdf-interest@w3.org
Message-ID: <op.som2cbxnw5l938@researchsft>

On Sat, 02 Apr 2005 13:52:11 +1000, Josh Sled <jsled@asynchronous.org>  

> URIs identify resources; the Accept header should serve only to
> negotiate the format of that resource, not to branch between different
> resources... you may want the HTML meta-data about the RDF data,
> someday. :)
> Why not have a URI for the resource, and a URI for the meta-data?
> GET /foo
> <foo>
>   <link rel="meta" href="/foo/meta" />
> </foo>

Because often the data you want about some resource isn't written by the  
person who happens to control what is served at that URI.

As a trivial example, W3C controls what is served at the URI associated  
with the RDF namespace. They don't happen to provide any RDF about  
human-friendly labels for the things defined there except in english.

As someone working primarily in spanish, I want to have spanish names for  
the various RDF Classes and Properties. There is no reason I cannot  
publish, somewhere on the Sidar site, these labels. (They're easy to  
produce...). But W3C doesn't necessarily know that I have done so. If I  
were the Mongolian Library, they are almost certain not to know that I  
have done so.

So querying W3C's server is of limited use.

The question then becomes, as Alistair noted, "so how do we find this  
stuff". I suspect the answer is the same as the answer to the equivalent  
question for the real web - we make use of search engines that go crawling  
around and providing a way of finding things we are looking for based on a  
large store of meta-information.

In the RDF case, I think the key information is about what stores can  
answer a given set of queries - I see the future search engines for the  
semantic web being based on query brokers that know where to get answers  
to a particular query, and how to distribute the query and consolidate the  
results. This relies on things like a query language (ideally a  
standardised one such as SPARQL, rather than two dozen different ones...),  



Charles McCathieNevile                      Fundacion Sidar
charles@sidar.org   +61 409 134 136    http://www.sidar.org
Received on Sunday, 3 April 2005 04:20:53 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 22:44:55 UTC