Re: Little Maps of the US

Still a bit foggy here.   A query would reference the dcat:distribution element 
for the timestamp but an inference of the dataset would follow along with it, 
yes?  And the dcat:distribution timestamp would reference when the dataset was 
distributed rather then when the actual event (eg: raining, snowing) occurred?
 
Michael A. Norton
 




________________________________
From: Gannon Dick <gannon_dick@yahoo.com>
To: Mike Norton <xsideofparadise@yahoo.com>
Cc: public-egov-ig@w3.org
Sent: Thu, August 26, 2010 8:53:57 PM
Subject: Re: Little Maps of the US

This relates to (sort of):
ISSUE-43 (abstract_dataset): attach specific properties to dcat:Distribution and 
not to dcat:Dataset [dcat]

although that is a more general case.

Two Mash-Ups of weather on geography, one for today and one for yesterday would 
be two dcat:Distribution or manifestations.  The common properties between the 
distributions would only be the geo properties which are fixed in time.  You 
could still figure out (where it rained|snowed|etc.)[Dataset] and 
(when)[Distribution date property].  This is the normal thing you do when 
designing a Data Base, but it's not always so obvious what is best.  If you 
wanted to get a list of noon day temperatures at a place for a year, another 
design (without the required 365 dcat:Distribution(s)) might be better.

There are always going to be query optimizations, but putting the meta 
properties in one place (attached to Distribution) is a best practice, I think.  
Make sense ?  


--- On Thu, 8/26/10, Mike Norton <xsideofparadise@yahoo.com> wrote:

From: Mike Norton <xsideofparadise@yahoo.com>
Subject: Re: Little Maps of the US
To: "Gannon Dick" <gannon_dick@yahoo.com>
Cc: public-egov-ig@w3.org
Date: Thursday, August 26, 2010, 8:49 PM

Gannon: "IMHO, the RDFa meta data should not include the 'Weather' parameters 
themselves."
 Why not?
Michael A. Norton 

From: Gannon Dick
<gannon_dick@yahoo.com>
To: Mike Norton <xsideofparadise@yahoo.com>
Cc: public-egov-ig@w3.org
Sent: Thu, August 26, 2010 1:11:50 PM
Subject: Re: Little Maps of the US


--- On Wed, 8/25/10, Mike Norton <xsideofparadise@yahoo.com> wrote:
Is this anywhere near the kind of data NWS utilizes in mapping meteorological 
data?

Yes, as a matter of fact ...

But let me back up just a little: The "Little Maps" are a grid point subset.  
The meta data is assumed constant, and it changes so slowly that that assumption 
is valid.  Weather observations and Forecasts are a time dependent Mash-Up. This 
topic is a huge subject of controversy at the moment, because a person's 
location is a similar type of Mash-Up. This debate runs counter to the 
Scientific Method since there is no reason to believe that observations are 
reproducible and computable with sound semantic principles.  This sort of 
thinking will get us both burned at the stake in Advertising Departments, Mr. 
Norton :o)  


The NWS has a beta
test of XML available, as well as some XSL transforms (which I have not had a 
good look at).  It would not be too difficult to add a Weather link to the 
"Little Map".  IMHO, the RDFa meta data should not include the "Weather" 
parameters themselves.

<http://www.weather.gov/alerts-beta/>


-- Gannon