Semantic Web URIs for Postal Codes

19 views
Skip to first unread message

Jamey Wood

unread,
May 26, 2009, 10:04:20 PM5/26/09
to GeoNames
One useful aspect of the GeoNames service is that it provides nice
semantic web URIs for places, e.g. the United States is:

http://sws.geonames.org/6252001/

Why does GeoNames note do the same for postal codes (e.g. 90210)?
Clearly, it does have information covering various postal codes, but
it does not (to my knowledge) provide a unique URI representing each
such postal code.

Thanks,
Jamey

RsH

unread,
May 27, 2009, 6:18:27 AM5/27/09
to geon...@googlegroups.com

Where would you stop, if you did this? You would then have to do the
same thing for Canadian Postal Codes, and for the codes used in the
U.K. and elsewhere in the world. After all, if GeoNames does that
for U.S. Zip Codes, or German Postal Codes, it has to do it around
the world. As the Canadian Postal Code can represent only part of a
building [Large office towers in Toronto can have a number of postal
codes, assigned to individual corporations resident in the building
with one code to cover the smaller occupants.] how do you cover that
sort of situation as well. Do you have several URI entries for that
one building? The building I live in is M2N 6G8. The building I used
to live in is M2R 1Y9. Those codes are unique to those specific
buildings, not shared with other resident locations. Even when a
postal code is shared, in Canada, it normally only covers a maximum
of about 100 homes... even if scattered over hundreds of square
kilometres... such as in Nunavut. Anyway, I am curious as to how
GeoNames could handle that sort of issue... as there could be
literally millions of unique postal codes, with overlap as you cross
geographical boundaries. For example, what prevents another nation
from also having an M2N 6G8 as a postal code, and if there are
indeed two or more M2N 6G8 postal codes in the world, how is one
distinguished from another?

FWIW
RsH
=======================================================
R. S. (Bob) Heuman <robert...@alumni.monmouth.edu>
Copyright retained. My opinions - no one else's...
If this is illegal where you are, do not read it!

Retention of this message in violation of Canadian
Privacy Laws will be prosecuted.

Jamey Wood

unread,
May 28, 2009, 10:44:57 AM5/28/09
to geon...@googlegroups.com
Hi Bob,

On Wed, May 27, 2009 at 4:18 AM, RsH <robert...@alumni.monmouth.edu> wrote:

On Tue, 26 May 2009 19:04:20 -0700 (PDT), you wrote:

>Why does GeoNames note do the same for postal codes (e.g. 90210)?
>Clearly, it does have information covering various postal codes, but
>it does not (to my knowledge) provide a unique URI representing each
>such postal code.
>
>Thanks,
>Jamey

Where would you stop, if you did this? You would then have to do the
same thing for Canadian Postal Codes, and for the codes used in the
U.K. and elsewhere in the world. After all, if GeoNames does that
for U.S. Zip Codes, or German Postal Codes, it has to do it around
the world.

Yes.  Personally, I do think it'd be good for GeoNames to provide unique URIs for postal codes in all countries possible.
 
As the Canadian Postal Code can represent only part of a
building [Large office towers in Toronto can have a number of postal
codes, assigned to individual corporations resident in the building
with one code to cover the smaller occupants.] how do you cover that
sort of situation as well. Do you have several URI entries for that
one building? The building I live in is M2N 6G8. The building I used
to live in is M2R 1Y9. Those codes are unique to those specific
buildings, not shared with other resident locations. Even when a
postal code is shared, in Canada, it normally only covers a maximum
of about 100 homes... even if scattered over hundreds of square
kilometres... such as in Nunavut. Anyway, I am curious as to how
GeoNames could handle that sort of issue... as there could be
literally millions of unique postal codes, with overlap as you cross
geographical boundaries.

Yes, I think those are all valid points about the challenges of describing postal code areas.  But I think they can be overcome.  Even if a postal code is small or sparsely populated (as in the examples you noted), why shouldn't it have its own unique URI?  And I can't imagine that postal codes would be the first case where multiple GeoNames places (of the same type) may have overlapping areas.
 
For example, what prevents another nation
from also having an M2N 6G8 as a postal code, and if there are
indeed two or more M2N 6G8 postal codes in the world, how is one
distinguished from another?

That kind of disambiguation is exactly where I would think that GeoNames URIs could add value.  How does one distinguish between "normal" places that have the same name (e.g. "Las Vegas, NV" and "Las Vegas, NM")?  One nice way is by using the unique URIs which GeoNames provides:

  http://sws.geonames.org/5506956/ (Las Vegas, NV)
  http://sws.geonames.org/5475433/ (Las Vegas, NM)

I would love to have the same for postal codes.

--Jamey

Francisco J. Lopez-Pellicer

unread,
May 27, 2009, 5:45:29 PM5/27/09
to geon...@googlegroups.com
> Anyway, I am curious as to how
> GeoNames could handle that sort of issue... as there could be
> literally millions of unique postal codes, with overlap as you cross
> geographical boundaries. For example, what prevents another nation
> from also having an M2N 6G8 as a postal code, and if there are
> indeed two or more M2N 6G8 postal codes in the world, how is one
> distinguished from another?

a) Do not use the postal code in the URI as Yahoo GeoPlane does. Then we
have an URI like

http://sws.geonames.org/231233122/

b) Use the postal code in the URI but prefixed with the ISO 3166 country
code. Then we have and URI like

http://sws.geonames.org/ca/m2n6g8/


-- javier


Marc Wick

unread,
May 30, 2009, 11:19:58 AM5/30/09
to geon...@googlegroups.com
Country and postal code are not sufficient to make it unique in all
countries. The place name would also be needed in the url and then the
difficulties with different spelling begin.
So I think a) is the cleaner url.

Marc

Marc Wick

unread,
May 30, 2009, 11:26:41 AM5/30/09
to geon...@googlegroups.com
I agree it would be nice and useful to have unique ids for
postalcode/places. A lot of users are actually asking for it. Not only
for semantic web applications, traditional rdbms applications have the
same desire for unique keys.

The problem is that it is not easy to maintain the keys and guarantee
that they do not change when importing new input data. At the moment it
seems better to allocate the resources on integrating the postal codes
into the main toponym database. With this integration (as alternate
names) they will automatically get an id. This id is not yet available
as semantic web URI, but when the id is available in the database then
it is a small step to publish it as rdf or other document.

Best

Marc
Reply all
Reply to author
Forward
0 new messages