RE: Question about namespace

Hi, 

Initially I had a misunderstanding that the namespace for an ontology is 
the physical URL to the ontology. It is not. 

If you want to make sure that the concepts in your ontology are what you 
intended, you can use <owl:imports> tag in your ontology to actually import  
the other ontology specifically from a physical URL. (See an example at
  http://www.flacp.fujitsulabs.com/tce/ontologies/2004/03/project.owl
which imports object.owl.)

But I still think it is not a bad idea to use the physical URL for the
ontology file as the namespace for the ontology when it is possible
and if there is no problem/confusion.

Regards,

Ryu

-----Original Message-----
From: public-sws-ig-request@w3.org [mailto:public-sws-ig-request@w3.org] On Behalf Of Bijan Parsia
Sent: Friday, August 26, 2005 12:23 PM
To: Tatiana Vieira
Cc: David Martin; public-sws-ig@w3.org; chen Xiaoyan
Subject: Re: Question about namespace


On Aug 26, 2005, at 12:02 PM, Tatiana Vieira wrote:

> Hi people,
>  
> As the topic here is namespace, let me ask a question. If a namespace 
> isn't necessarily an URL

It isn't. See URNs.

>  (it isn't necessary to be a physical location), how can anyone 
> discover the concepts declared in an ontology?

Ontology != namespace. An ontology (thought of as a logical theory) can 
have axioms using terms from an arbitrary number of namespaces 
(actually, there *is no such thing* as a namespace in RDF or OWL; there 
are only URIs; the closest approximation to a "namespace" is "uris 
sharing a common prefix"; but note that all http uris share some common 
prefixes!)

Terms can be defined in more than one ontology.

>  I mean, how can I use a specicified vocabulary defined in any place 
> in Web if I can't find it?

You can find it some other way, e.g., by checking a list, or by being 
emailed the ontology, etc. It's not *that* hard.

> And, also, how can we be sure that namespaces will not conflict,

You cannot be sure that people will use the term the same (or in 
compatible) ways.

> I mean, that two diferent users will not put an igual namespace for 
> different documents?

They might even put contradictory axioms for the very same term! (which 
is more severe, obviously).

C'est la vie, eh? The key for a consumer is to determine which 
documents they trust. Some people take it as a strong rule that you 
should trust the documents retrievable from the URI owner's server wrt 
those URIs. Others (like me) are not so restrictive.

Cheers,
Bijan.

Received on Friday, 26 August 2005 16:45:28 UTC