Re: URI for language identifiers

Patrick.Stickler@nokia.com wrote:
> 
> > -----Original Message-----
> > From: ext Jan Algermissen [mailto:algermissen@acm.org]
> > Sent: 02 April, 2003 10:46
> > To: Stickler Patrick (NMP/Tampere)
> > Cc: www-rdf-interest@w3.org
> > Subject: Re: URI for language identifiers
> >
> >
> > Patrick.Stickler@nokia.com wrote:
> >
> > >
> > > But if the URI denotes two things, how do you differentiate
> > > between statements made about one versus the other?
> >
> > The URI does not denote two things. There are just two kinds
> > of properties
> > on topics that use URIs as values. The semantics of the properties are
> > different.
> >
> > The whole thing is different because in topic maps, you have
> > an unlimited
> > number of possibilities to identify what a given topic represents.
> 
> Are you denoting topics using URIs or not? 

TMs use property/value pairs to identify what they represent. The URI
allown is simply not enough.

If you are, and that
> topic can represent more than one thing, then that's ambiguous.

A topic never represents more than one thing.

[..]

> > > > The denotation is not ambigous but there are two *ways to
> > use* of URIs
> > >
> > > If a single URI can denote both a web resource and an
> > abstract subject,
> > > then it is ambiguous. Period.
> >
> > A URI allways denotes the 'web page'. Period. ;-)
> 
> Then TM's are completely unusable for most of what I do if I
> can't use a URI in a TM to denote an abstract concept.

What you use is a propery/value pair where the URI happens to
be the value.
> 
> > > If the interpretation of a give URI depends
> > > on context, then it is ambiguous. Period.
> >
> > The context is explicit, so what is ambigous about that?
> 
> Where is the context explicit? If all I have is a URI, and
> nothing else, how to I determine which of the many possible
> things it might denote in a TM?

You have more than the URI, you have the property also. Example:

topic1 [SubjectIndicators: (http://www.w3.org/ , http://www.w3.org/Consortium)]

This topic has two subject indicators (the two resources addressed by the URIs)
and (if they are well chosen subject indicators) , rendering the resources for
human perception will tell you what the subject of the topic is.

topic2 [SubjectAddress: http://www.w3.org/Consortium]

This topic is a surrogate for the information resource living at
http://www.w3.org/Consortium .

This is not ambigous, because the semnatics of SubjectIndicators and
SubjectAddress are well defined.

[Aside: XTM doesn't take the resource/information resource indirection
 into account, it is assumed that http://www.w3.org/Consortium addresses
 the bits and bytes. Which is wrong but OTH a (desired) limitation of
 REST. If you write  http://www.google.com dc:language "English" you have
 the same problem (due to conneg), so it's not really only an XTM problem
 I think]

> 
> If I ask some knowledge base "Tell me about the thing denoted
> by this URI" what is it going to tell me? 

You ask the TM knowldege base either: what is the thing indicated
by this URI or what is the thing addressed by this URI. Simple ;-)

Is it going to
> describe all the possible things it could denote? The web
> page, the subject, whatever?
> 
> URIs should have consistent, *global* (meaning context-independent)
> and unambiguous denotation. Period.

As long as topic maps avoid using the URIs that do NOT denote
'web pages' it's fine since in XTM land it is assumed that a
URI allways denotes a web page.

> 
> If XTM doesn't provide that, then it is inherently incompatible
> with RDF.

Well, as long as the author of a topic map can find out what URIs
don't denote information resources I don't see a problem here. The
thing is that RDF aims to integrate with the Web while X(TMs) merely
aim to use it. What I learned in the last months though is that there
are bad ways to use the Web Architecture (e.g. careless use of fragment
identifiers) and that current XTM practice (the documents out there)
could do better.

Jan

> 
> > > XTM has ambiguous use of URIs.
> >
> > No. Period. ;-)
> 
> Sorry. You seem to be mistaken here. If you understand what
> "globally unambiguous" means.
> 
> XTM does not appear to use URIs in a globally unambiguous manner.
> 
> 
> Patrick
> 
> --
> Patrick Stickler, Nokia/Finland, (+358 40) 801 9690, patrick.stickler@nokia.com
> 

-- 
Jan Algermissen                           http://www.topicmapping.com
Consultant & Programmer	                  http://www.gooseworks.org

Received on Wednesday, 2 April 2003 04:02:40 UTC