[Fwd: query about best practices for URI resolution]

In case you missed this.
Guus

-------- Original Message --------
Subject: query about best practices for URI resolution
Resent-Date: Tue, 21 Oct 2008 06:41:52 +0000
Resent-From: semantic-web@w3.org
Date: Mon, 20 Oct 2008 14:17:04 -0700
From: John Graybeal <graybeal@mbari.org>
To: semantic-web@w3.org


All,

Background: The Marine Metadata Interoperability project is about to
deploy a server of ontologies, particularly containing vocabularies
and mappings from marine science.  (Initially most of the ontologies
will be fairly simple vocabularies and mappings, but over time an
increasing number will be more complex.) We intend to provide human-
centric resolvable URIs representing the resources of both the served
vocabularies and their terms.  (We appreciate that the URIs should be
considered 'just strings', not semantically parseable, but that
semantic information can be used to uniquely build a URI.)

WIth the publication of http://www.w3.org/TR/swbp-vocab-pub/  (thank
you!!!), we now have just a few relatively narrow questions.

1) Some members of the community we serve will want to use URLs for
their terms, and some will want to use URNs, depending on
circumstances; sooner or later, some of the same resources will need
to be accessed via both forms, so we will want to resolve both.  Is
the best approach: (a) make one form the URI of the resource, and the
other one just another link or URI that we resolve back to that
resource, (b) same as (a) but we create an independent set of
resources so that we can explicitly relate the alternate resource to
the original, or (c) allow either form to be used as 'the resource',
and use owl:sameAs to relate them? Or is there another agreed
approach?  The last seems best because it make the relationship
explicit for inferencing, but I figure a lot of purists get upset
about two URIs for the same resource.

2) Is there a de facto or best practice for specifying URNs between
the time that you know what your URNs should look like, and the time
your application has not yet been approved?  I've seen a lot of
different practices used or recommended....

3) Is it generally OK to serve RDF (e.g., for a term) as a complete
OWL XML file -- should semantic web applications that expect RDF be
able to process that?

I appreciate your considered advice, or references to documents that
provide a credible answer.

John


--------------
John Graybeal   <mailto:graybeal@mbari.org>  -- 831-775-1956
Monterey Bay Aquarium Research Institute
Marine Metadata Interoperability Project: http://marinemetadata.org