Re: Not needed? Distributed URI Discovery

Hi Max,
everybody

Es begab sich aber zu der Zeit 23.03.2005 21:32,  da Max Voelkel schrieb:

>>>At present, there is no formal, generalized mechanism whereby a Web Agent,
>>>upon discovery of a URI, and lacking knowledge about that URI, can query the
>>>Originator of the URI in order to obtain an RDF description of the URI.
>>>      
>>>
>
>Lets compare URIs with symbols. When I discover a new term, I can not
>ask the term, what it means. I ask a knowledge source (friends, books,
>search engine) about it.
>  
>
wrong. You can ask the term.

http://www.kwark.org/Gfx/2005/2005Week11/dscn9703.Pijkestraat.jpg.html
click on "rat"
http://www.kwark.org/x/person.pl?q=http%3A%2F%2Fxmlns.com%2Fwordnet%2F1.6%2FRat

and then enter
http://xmlns.com/wordnet/1.6/Rat

and you are there.

the term is the uri is the word is the resource is the graph.
And that is much easier and I do not get a headache with "intermediate" 
systems i have to use to
* get the uri
* get the rdf
* know where to query for it
etc


>Why do we have to make things different on the web? 
>
see above. we already did it.

>When I find
>a RDF document with URIs I don't know i just ignore them. If the RDF
>document was well-written it should contain rdf:seeAlso links to URLs
>of RDF-documents describingthe terms. Mights this be a solution?
>  
>
it is, in files like "about-me foaf files" (i.e. 
http://www.leobard.net/foaf.xml) where I cannot identify a person by a 
uri because the person has hundreds of uris. so I use an anonymous 
resource, tag it with a owl:IFP (inverse functional property, sha1sum of 
mailbox) and then add,
if known, the RDFS:seealso

but for identifyable things (like rss etc) you MUST use the URL of the 
resource and everybody is happy.

>It is inspired by the WWW approach of links and by the design
>criterion of separation between identity (URI) and location (URL). I
>think also "_:1 rdf:type foo:isCrawlable" or similiar would be
>helpful.
>  
>
it does not help
see
http://www.w3.org/DesignIssues/HTTP-URI.html

and the "Exercises" at the bottom.
for them, we would have to do many more hacks.

>My conclusion is thus we need no index.rdf, no URI originator, no MGET etc.
>__ Location is not identity. __
>  
>
thats not true in today's semantic web.
all the good tools use uris as identity or owl:IFP's to identify stuff 
and a central/distributed index.

You must not close your eyes against the killer applications:
* RSS - over a zagillion websites using it - most of the weblogs and 
news sites
   items are identified by URL 
  example:
   http://leobard.twoday.net/index.rdf

* foaf codepiction - many many pics out there - the URL of the jpg is 
the resource identifier
   http://swordfish.rdfweb.org/discovery/2001/08/codepict/

* Wordnet - the term is the URL is the identifier
    http://xmlns.com/wordnet/1.6/Identifier


>
>Ok, what if somebody else adds statements about a URI? Well, then i
>either need something like
>a) a search engine = centralized infrastructure or
>b) something like traceback = distributed, networked infrastructure
>   -> we need a standard for RDF-traceback-servers!
>
>  
>
weblog traceback/pingback (or whats it called again...?)

and: that's how the web works:
GOOOOOOOGLE
has the centralized infrastructure
(and yahoo and msn and .....)
they have the money to do it, so let them do it. fine with me,
as long as there are more than one.


cheers
Leo

btw,
the good old uri crisis!
we love it, every 3 months it comes up again :-)
(I mentioned it to you in the talk i gave in karlsruhe back in november :-)