The statement that "the mechanics have been worked out" presumes a lot  
of things that are not always true or are often not properly practiced  
(not all browsers behave properly, it works only for URLs and not for  
URNs, a proper style sheet has to be created for proper  
presentation, ...), and if using the '#' format for term URIs, forces  
the download of (potentially very large) files just to see one piece  
of data. And in the semantic web community there is still quite a bit  
of discussion about whether the redirect approach is best.

Similarly, the wisdom of inclusion of implicit information within the  
URI is hotly contested in the semantic community.  (A large percentage  
believe URIs should be opaque, though MMI agreed with Nan that  
semantic URIs are best in some circumstances [1].)

Our own conclusions and references about the best URI approach for the  
scientific community are documented on-line [2,3] and we welcome  
critiques (and expect dissenting opinions, if the semantic community  
discussions are any indication!).

But I wonder if these topics are best discussed elsewhere, like in the [log in to unmask] 
  mail list?  I see the advantage of discussing them in the scientific  
community context, but such a discussion can get very far down into  
the weeds indeed.

John

[1] Constructing URIs: http://marinemetadata.org/apguides/ontprovidersguide/ontguideconstructinguris
[2] MMI Ontology Providers Guide: http://marinemetadata.org/apguides/ontprovidersguide/ontguideconstructinguris
[3] MMI Semantic Framework Concept: http://marinemetadata.org/semanticframeworkconcept


On Feb 25, 2009, at 9:01 AM, Thomas Baker wrote:

> On Wed, Feb 25, 2009 at 10:53:29AM -0500, Joe Hourcle wrote:
>> The only problem that I have with URIs is when they don't resolve  
>> so that
>> we can get information about what it means.  (and well, that one  
>> doesn't,
>> as lcsh.info had to shut down).  Without resolution, URIs are less  
>> useful
>> than strings.
>
> The mechanics for resolving URIs to useful information -- and
> in accordance with current principles of Web architecture --
> have been worked out as described in [1] and [2].
>
> If an Apache Web server is configured accordingly, a
> property URI such as
>
>    http://www.w3.org/2008/05/skos#broader
>
> will resolve either to a Web page (if you are a human clicking on
> the URI in a browser):
>
>    http://www.w3.org/TR/2008/WD-skos-reference-20080829/skos.html#broader
>
> or to a machine-processable RDF schema (if you are requesting
> the information via an RDF application):
>
>    http://www.w3.org/TR/2008/WD-skos-reference-20080829/skos.rdf
>
> The redirects happen automatically.  If you have access to
> a computer running the command-line utility "curl", you can
> examine the intermediate steps, as attached below.
>
> Tom
>
> [1] http://www.w3.org/TR/cooluris/
> [2] http://www.w3.org/TR/swbp-vocab-pub/



> From: Nan Galbraith <[log in to unmask]>
> Date: February 25, 2009 8:30:58 AM PST
> To: [log in to unmask]
> Subject: Re: Identifying controlled-vocabulary terms using URIs
> Reply-To: DCMI Science and Metadata Community <[log in to unmask] 
> >
>
> Hello all -
>
>> The only problem that I have with URIs is when they don't resolve so
>> that we can get information about what it means.  (and well, that one
>> doesn't, as lcsh.info had to shut down).  Without resolution, URIs
>> are less useful than strings.
>
> It would be really useful if a standard for URI syntax could be  
> established
> to make them implicitly provide human-usable information.  Using terms
> that have meaning lowers the error rate and makes the metadata much
> more valuable and resilient.
>
> In the example below, dc:subject "World Wide Web" is at least useful  
> to
> someone reading metadata, while
> dc:subject "http://lcsh.info/95000541#concept" is meaningless. How  
> much
> more  useful this would be if the URI were something like
> "http://lcsh.info/terms#worldwideweb"  -  especially (but not only)  
> since
> the lcsh server is offline.
>
> We use the Climate and Forecast NetCDF implementation for our  
> meteorological
> and oceanographic data. The single most useful characteristic of CF  
> is that it uses
> standard names which are human-readable. The fact that these terms  
> have
> precise definitions available on line is important, but without the  
> meaningful names
> we'd have many more errors in our documentation.
>
> Nan

John

--------------
John Graybeal   <mailto:[log in to unmask]>  -- 831-775-1956
Monterey Bay Aquarium Research Institute
Marine Metadata Interoperability Project: http://marinemetadata.org