Using tag: scheme to identify resources?

What is the general (i.e. your) opinion on using the "tag:" URI scheme
<http://taguri.org/> to identify resources?

The first issue is the usefulness of the scheme.  Some people in my group
expressed concern that they weren't an officially endorsed scheme.  It is my
understanding that "tag:" is a RFC and also officially recognized by the IETF.

The second problem come from this advice from the web site:

] As with XML Namespaces, there are two schools of thought. Do you want 
] the web's default mechanism at your disposal for fetching information 
] about the identified thing? If so, then HTTP URIs are probably better 
] used.

It seems to me that from RDF's POV, it seems that there is no preference since
names are opaque (no RDF information of consequence can be determined from just
the name.  You need at least a complete statement.  Yet from a software
application's POV, it may try to dereference a URI to get more information. 
Also from the graph modeler's POV, it is important to have authority over your
namespace (or else there could be graphs you can't safely merge with).  Are
these issues relevant?  Anything I'm missing?

For ontology schemes, it is advantageous to share them with others so putting
them on the Internet is a good thing.  That's why a "http:" or "ftp:" URI is
useful.  Is there a reason to prefer "http:" over "ftp:"?  What should you do
if you provide the ontology in several formats (RDF/XML, N3, other)?  Each
would be located at a different URI in that case.  It is hard to engineer
serving different media types if you don't have access to your server or don't
want to change its configuration.

For instance things which are not part of an ontology scheme, I think "tag:"
names work well.  When talking about non-computer entities like people, it is
hard to enforce a permanent canonical name.  If I use an email address or web
page location, that could still change at any moment (it certainly has for
me!).

Names based on "tag:" add a temporal component that can cheaply persist
pointing to the same authority.  That is, the owner of "tag:example.com,1999:"
in 1999 can still the owner of "tag:example.com,1999:" in 2005 without having
to pay anything.  That simplifies the issue of who has authority over the
domain (or in tag's case: the owner of the domain at a specific date).

Are the following rules sensible?

1) If you do not have permanent control over the domain, get temporary
permission and use a "tag:" URI.  If you can't get that, use a "urn:uuid:"
name.

2) If it doesn't make sense to dereference your resource, use "tag:".

3) If it doesn't have a stable location over time, use "tag:".  This is why
"file:" makes a poor choice of a scheme for resources shared with other people.

4) If it is an ontology scheme, consider using a "http:' name.  Otherwise,
choose the scheme that makes the most sense for that domain.

Any other tips for assigning names?



--
Jimmy Cerra
https://nemo.dev.java.net


		
__________________________________ 
Discover Yahoo! 
Stay in touch with email, IM, photo sharing and more. Check it out! 
http://discover.yahoo.com/stayintouch.html

Received on Thursday, 9 June 2005 06:46:42 UTC