Re: Re: Addressing in the SW!

Peter Vojnisek <et@progos.hu> schrieb am 26.06.2003, 15:47:58:
> Jan,
> 
> Thank you for the information. I understand your point, but I think
> http://www.example.org/persons/Joe/Name has two problems from my point 
> of view:
> 
> 1. If I know well the standard we have to address Joe like 
> http://www.example.org/persons#Joe and not 
> http://www.example.org/persons/Joe I have already tried this with Jena2, 
> and a nice NullpointerException was the result... :-(
> There are some examples with / at http://purl.org/dc/element/1.1 and 
> also fails with Jena2. In this case is DC W3C standard?

Besides from any requirements that RDF itself or RDF software imposes on
how URIs can be composed, I think that using fragment identifiers in
general are problematic:

Suppose you really use

  http://www.example.org/persons#Joe

to address Joe, the I'd naturally expect that a GET on this URL will
return a representation of Joe. The problem is, that when typing the URL
into the Browser's location bar, the browser will send
only

  http://www.example.org/persons

to the server, applying the '#Joe' portion to the result of that
request. In other words, the browser will receive the whole list of
persons only to display Joe's representation to me.

So, I'd try to avoid the fragment as much as possible.

> 
> 2. There is no information about the namespace of the name. Imagine if 
> Joe has two name properties: one from http://nick.org/person and the 
> other from http://seriouspeople.com/person namespace. How can we select 
> the one we want?

Well, you could use URLs like

http://www.example.org/persons/Joe/Name?namespace=http://seriouspeople.com/person
(ommited URL encoding of query string)

to address a particular name of Joe.

Jan

> 
> Solution:
> a) There is no way to address this simple question in one uri.
> b) It can be used within a namespace in non-standard way.
> 
> What do you think?
> 
> Peter
> 
> >
> >you could use URLs like the follwoing
> >
> >http://www.example.org/persons 
> >
> >    to address the list of all persons
> >
> >http://www.example.org/persons/Joe
> >
> >    to address Joe
> >
> >http://www.example.org/persons/Joe/Name
> >
> >    to address Joe's name
> >
> >http://www.example.org/persons/Joe/Age
> >
> >    to address Joe's age
> >
> >etc.
> >
> >Content negotiation between server and client could return the various
> >'entities' in diverse formats, among them "Joe" for
> >
> >http://www.example.org/persons/Joe/Name
> >
> >Does that help?
> >
> >Jan
> >Peter Vojnisek  schrieb am 26.06.2003, 14:43:38:
> >  
> >
> >>Jan,
> >>
> >>    
> >>
> >>>> If we have a class instance: http://example.org/persons#Joe
> >>>> and we know that there is has a property http://example.org/person#name
> >>>>
> >>>> Can we reference to Joe's name in an uri?
> >>>>   
> >>>>
> >>>>        
> >>>>
> >>>To make sure I understand you:
> >>>
> >>>You think of Joe's name as a concept in it's own right and you want to
> >>>make a statement about the concept, yes?
> >>>
> >>>As in:
> >>>
> >>>"(Peter) --likes--> (Joe's name)".
> >>>
> >>>is that what you want to do?
> >>>
> >>> 
> >>>
> >>>      
> >>>
> >>This idea is coming from that we wanted to address the property from 
> >>applications. Object and Datatype properties are the same from this 
> >>point of view, so an URI or a value will be the result of the query.
> >>
> >>Your example assumes that Joe's name is not a DatatypeProperty. But I 
> >>think you know what I mean.
> >>
> >>something is like this:
> >>Person - is a - Class
> >>name - is a - DatatypeProperty
> >>name - has a datatype of - string
> >>Person - has a property called - name
> >>---
> >>Joe - is a - Person
> >>Joe - has a name - "John"
> >>
> >>The problem is addressing Joe's name. The query result would be "John"
> >>
> >>    
> >>
> >>> 
> >>>
> >>>      
> >>>
> >>>>   I have seen this somewhere on the web: 
> >>>>http://example.org/person/Joe#name I don't think it is correct, we lost 
> >>>>the namespace of the predicate.
> >>>>
> >>>>   Do you know a sollution for this problem? Is there any W3C standard 
> >>>>for this kind of addressing?
> >>>>   
> >>>>
> >>>>        
> >>>>
-- 
Jan Algermissen                <algermissen@acm.org>
Consultant & Programmer

http://www.topicmapping.com
http://www.gooseworks.org

Received on Thursday, 26 June 2003 10:32:08 UTC