Re: SemWeb Non-Starter -- Distributed URI Discovery

Es begab sich aber zu der Zeit 01.04.2005 19:36,  da Miles, AJ 
(Alistair) schrieb:

>... right but, HTTP still provides a generic mechanism for retrieving an 'authoritative' representation of a resource (i.e. HTTP GET) - whether the resource exists or not does not diminish the requirement for such a mechanism.
>
>So the semantic web requires some generic mechanism whereby an agent can attempt to locate and query a set of RDF statements about the resource denoted by URI x that has been endorsed by some entity y - whether such statements exist or not similarly does not diminish the requirement.  N.B. In the most common scenario, y is the owner of x.  
>  
>
this general mechanism is already implemented and the thing I know best 
is uriqa

jup, so let us standardize uriqa or something else.

alas, standardization is always a tough thing to do, all these boring 
documents to write, etc.
DanB once said to me that the dublin core work was hard and tough work 
while making foaf was a joyful experience.
So standardization could work like it is now: we see that it works,
we hack code.
it spreads.


Uriqa does most of the stuff we need and anything else would probably go 
deeper to URI CRISIS

Example
one uri, ie a picture uri like
http://www.kanzaki.com/works/2004/imgdsc/040207_1656.jpg
may be described by RDF like
http://www.kanzaki.com/works/2004/imgdsc/040207_1656.rdf

tada,, it works.
but the semantics behind is tough: parse for a "." in the uri and 
replace it by ".rdf"
that does not work in general, we had the discussion already years ago.

so the MGET/ GET approach by URIQA in combination with known webservices
at /uriqa? is a very good thing to start.
If mget fails, add a config string with the uriqa service uri and you 
are done.

We used this to connect mindraider.sf.net to www.gnowsis.org
was a one day hack, done without much programmer communication,
as the standard is easy.

cheers
Leo

Received on Tuesday, 5 April 2005 12:08:10 UTC