[OSGeo-Discuss] RE: Representing Places With Intelligent URLs

[Geoff Hay]

Hi
The knowledge you are trying to encode should be represented as associations between individuals (this place contains that place etc) and concepts (city, park, post office delivery area, etc) (as in OWL) rather than a URI scheme (see Geonames).  The basic idea is to represent places in a way that allows inference (make implicit knowledge explicit) i.e. logical consequence 
e.g. 
Explicit: a country only has only one capital city
Explicit: NZ is a country 
Explicit: Wellington is the capital of NZ
Explicit: 'Te Upoko o te Ika a Maui'  is the capital of NZ
Implicit: Wellington and 'Te Upoko o te Ika a Maui' are the same place 

- you cant do this nicely with a URL scheme but an OWL reasoner can make such conclusions - yehar Semantic Web.  Actualy there is really no problem with your URI scheme otherwise. It looks exactly like what you would expect for REST Web Services URLs - as long as you don't expect your URLs to be the ultimate and final identifiers - that would break both of the two main assumptions behind the semantic web and its underlying formal logics.

regards
Geoff
________________________________________
From: discuss-bounces at lists.osgeo.org [discuss-bounces at lists.osgeo.org] On Behalf Of Landon Blake [lblake at ksninc.com]
Sent: Wednesday, 6 October 2010 12:45 p.m.
To: OSGeo Discussions
Subject: [OSGeo-Discuss] Representing Places With Intelligent URLs

A talk at the recent Location Business Summit and some reading I've done
about the semantic web and microformats lately got me to thinking about
a standard way to represent places, place names, place data on the web.
(I must admit I'm a desktop software guy, not a web programmer.)

I thought it would be awesome if there was a way to create a unique URL
for places that was somewhat intelligent to humans. If this URL could
point to a folder on a server with some basic information about a place,
that would be even better.

So I took a stab at creating this type of URL for my city, the City of
Stockton. Here it is:

http://www.standardwebmarkup.org/standard_places/north_america/united_st
ates_of_america/california/san_joaquin_county/city_of_stockton/

You can see the URL follows a logical hierarchy, and it would be easy to
determine what the URL for the City of Sacramento, San Joaquin County,
or Victory Park in the City of Stockton would be. Obviously the
continent/country/state/county/city/location URL pattern would have to
change for other parts of the world.

I put a very simple HTML file with data about the City of Stockton here:

http://www.standardwebmarkup.org/standard_places/north_america/united_st
ates_of_america/california/san_joaquin_county/city_of_stockton/info.html

The current info.html file is just a skeleton. It's more of a place
holder right now than anything else.

My thought was to also put a WKT file (place.wkt) representing the
location of the place and a simple text file (data.txt) with facts about
the place at this same URL:

http://www.standardwebmarkup.org/standard_places/north_america/united_st
ates_of_america/california/san_joaquin_county/city_of_stockton/

Now, if someone wanted to write content about the City of Stockton, they
could simply do something like this:

<a
href="http://www.standardwebmarkup.org/standard_places/north_america/uni
ted_states_of_america/california/san_joaquin_county/city_of_stockton/">S
tockton</a>

If everyone that was putting web content about Stockton online did the
same thing, search engine and other tools would be able to link data
from this web content to a single location.

This becomes even more powerful if we come up with some rules for the
content of the info.html file, place.wkt file, and the data text file.
Here are some examples:

(1) Specify that the place.wkt file have both a point and a polygon WKT
representation, or a linestring representation, of the place when
appropriate.

(2) Specify that the info.html file use a list with alternate place
names. This list would be identified with an html class value of
"alternate_place_names".

(3) Specify that the data.txt file contain a relationships section that
can contain an optional relationship in the form of: City is the County
Seat of County. (Stockton is the County Seat of San Joaquin County.)

(4) Standardize the way common place facts are stored in the data.txt
file. Population and area are examples.

I realize there are some problems with this overall scheme. How do you
store a city that straddles a state boundary, for example? Or what if
you want to have a URL for the location of the Pacific Garbage Patch?

However, I think we could use this system to uniquely identify and
describe a lot of places in the world. We could then work on how to
handle the edge cases.

Is anyone else interested in ironing out the kinks for a system like
this? Is there already a system like this in place? (If so, I have just
revealed my great ignorance to everyone on this mailing list.)

I'm interested in setting something up that could be maintained by a
group of geospatial professionals, and not by any one company.

I'm not sure how this system I describe would tie in with geonames. My
first reaction when I stumbled on geonames is I couldn't find a unique
and human understandable URL for a place.

Still, I'm interested in microformats and place names, and I'd like to
see a system like this that was "open" and non-proprietary.

Let me know what you think.

The Sunburned Surveyor




Warning:
Information provided via electronic media is not guaranteed against defects including translation and transmission errors. If the reader is not the intended recipient, you are hereby notified that any dissemination, distribution or copying of this communication is strictly prohibited. If you have received this information in error, please notify the sender immediately.
_______________________________________________
Discuss mailing list
Discuss at lists.osgeo.org
http://lists.osgeo.org/mailman/listinfo/discuss