W3C home > Mailing lists > Public > public-lod@w3.org > March 2009

Re: GeoNames and Spatial Queries

From: Peter DeVries <pete.devries@gmail.com>
Date: Thu, 5 Mar 2009 23:51:11 -0600
Message-ID: <3833bf630903052151k271f936dvb87a062b7829d4f1@mail.gmail.com>
To: Christopher St John <ckstjohn@gmail.com>, public-lod@w3.org
This might not be of specific interest for your specific project, but may be
to others ...
There are climate and biogeographical data sets that have a 1km x 1km
resolution.

Specifically, WorldClim http://www.worldclim.org/

*"WorldClim is a set of global climate layers (climate grids) with a spatial
resolution of a square kilometer."*

It might make sense to map these to uri's, so they can be easily queried.
Also other data can then be tied
to those uri's. One difference between using something like this and
Geonames is the WorldClim data
set is made up of standard  1km x 1km squares for the entire planet (There
may not be records for the areas near
the poles)

I have been thinking about doing something like this but it might be better
to have a group develop some well
thought out and widely adopted standard.

I have been using GeoNames to tie the expected and observed status for
species. This allows you to ask
if a particular species is expected or has been observed in a state our
county. I have a specific meaning
for "Expected" at that is that you would not be completely surprised to
collect or observe it. A wild Tiger would be be
Unexpected for Wisconsin, a wild Cougar would be Expected (although very
rare)

Most of the observation data is not currently publicly available, but you
can get some sample observations at:

http://rdf.geospecies.org/observations/index.rdf

One thing to note, this particular structure is a little unstable since I
have been trying modify it to work better with another
existing vocabulary. The GeoNames part however, is relatively stable.

Also the individual species uri's, for those species we know something
about, also  give return data on Expected status, currently for Wisconsin
and Iowa. (we have little information on Iowa species other than mosquitoes)

Also, I am open to modifying the structure for observations if anyone has
comments or suggestions.

- Pete





On Tue, Feb 24, 2009 at 10:30 AM, Christopher St John <ckstjohn@gmail.com>wrote:

> I'm looking for feedback/pointers on best practices
> for finding objects in the Linked Data cloud given
> a geographic area of interest.
>
> Tom Heath's excellent Linked Data tutorial in Austin
> last week inspired me to do a quick Linked
> Data-based iPhone application. Think
> DBpediaMobile[1], only with a very different user
> interface.
>
> I spent some time researching the topic, but I was
> having a hard time figuring out what the general
> consensus was (or if there was one yet) I'd be happy
> to summarize responses into a FAQ answer.
>
> The DBpediaMobile paper[2] says:
>
>  "The map view is built from RDF triples
>  obtained by sending the currently visible area
>  ... to the server, where they are rewritten as
>  a SPARQL query and issued to a Virtuoso server
>  that hosts DBpedia’s geocoordinates..."
>
> DBpediaMobile uses GeoNames data, and geonames has
> an API with a query called "findNearby" that looks
> promising, but I'm assuming that's "cheating".
> Calling out to an API breaks the idea of linking
> within the data, and means that you can't browse
> through without special integration code.
>
> There are proposed extensions to SPARQL to handle
> spatial semantics[3]. I suspect that would solve the
> "cheating" issue (because the query would presumably
> be generic enough to work with any possible data
> source), but GeoNames doesn't appear to handle any
> SPARQL at all. (I think)
>
> But the excerpt form the paper indicates that the
> public GeoNames database is not being used. Instead
> the data has been loaded into a private datastore,
> presumably one that supports special spatial SPARQL
> queries?
>
> Is that the case? I can always just load up whatever
> data I need into PostgreSQL (which has excellent
> geodata support) and drop down into SQL queries, but
> that seems against the spirit of the thing. And of
> course at that point it's not really Linked Data at
> all because it's not on the web, or shared, or RDF.
>
> Feedback/hints/obvious-things-I-missed/
> corrections-to-misapprehensions greatly appreciated.
>
>
> -cks
>
>
> [1] http://wiki.dbpedia.org/DBpediaMobile
> actually seems to work pretty well on an iPhone,
> but there's no GPS info and i'd like to see a
> native app instead of something running in the
> browser. And what I have in mind has a totally
> different UI.
>
> [2] http://beckr.org/wp-content/uploads/DBpediaMobile.pdf
>
> [3]
> http://data.semanticweb.org/workshop/terra_cognita/2008/paper/main/1/html
>
>
> --
> Christopher St. John
> http://praxisbridge.com
>
>


-- 
---------------------------------------------------------------
Pete DeVries
Department of Entomology
University of Wisconsin - Madison
445 Russell Laboratories
1630 Linden Drive
Madison, WI 53706
------------------------------------------------------------
Received on Friday, 6 March 2009 15:03:55 UTC

This archive was generated by hypermail 2.3.1 : Sunday, 31 March 2013 14:24:19 UTC