Fwd: RE: [Geowanking] intro

[Daniel Smith]

At 4:51 PM -0500 4/2/03, Joshua Schachter wrote:
>I still get the impression that you shouldn't alias a file in order to
>indicate metadata.
>
>How do you extend this to, say, authorship?

I wouldn't...

It could be something that gets tucked away in RDF.
I'm not sufficiently up to speed on things RDF, but my
impression is that it could work.

My thought was to minimize extra components that end
up in a URL (or anything that I would call "ThereThen")
There are tons of methods to express
author, keywords, and all of our other favorite
info in metadata.  But... Location and Time
really can be narrowed down to a couple of
pretty universal forms [1]  I can't think
of any other metadata that's as easy to put
into a consistent form....ok, temperature
and its cousins: humidity, etc.  The form
of everything else is too open to dispute :-)

I like the GeoURL ICBM tag.
I like the RDF produced by
Moveable Type: dc:date="2003-03-29T21:48:23-08:00".
The two of them together are equivalent
to a ThereThen address.

Someone could add to the scheme on a per-domain
basis - but, every component added drastically reduces
the ease of getting "buy in" from potential
tool makers and end-users :-)

I'm thinking along the lines of the X11 Consortium
"mechanism, not policy" mantra.  Someone could produce
or resolve an address any way they want, or add on to it
within their organization.  The constant is that
the ThereThen portion of a URL (RDF or otherwise)
be consistent and portable.  It's deliberately
simple.

As for the whole bit about URL aliasing, I'm
thinking about all of the docs out there
now.  You can't go back and change them, but
you can derive additional meaning from them
so that future searches can hit/not hit them.
What other way could be used to derive and
store the same info, and would it be as
easy to use/universal as a URL?
For docs going forward, perhaps it is sufficient
for many folks just to tuck away TT functionality
in an RDF.  Others may want the handiness of
being able to drag and drop a group of TT
addresses into some sort of graphical resolver...

This is all a bit off-the-cuff (and I am starving
at the moment, but want to get this out), so I
hope it makes some sense.  cheers!


Daniel


[1] I looked at the dgeo metadata at:
http://www.dgeo.org/metadata.shtml ...
Cells, and... domains, and.. no emphasis on
time... it looks like something that was overly
ambitious.

I sure wouldn't propose a .tt top level domain,
but I could see having tt.example.com subdomains,
which are hit on by web services seeking to
resolve TT addresses.

-- 
Daniel L. Smith - Sonoma County, CA - AIM: SonomaDaniel