[OSM-talk] postcode idea
Dan.Putler at sauder.ubc.ca
Thu Mar 30 19:22:35 BST 2006
It turns out that you don't want the street number of every house and
business in an area (which diminishes the privacy concerns). Instead
you want a more limited set of information. A standard road network
file contains (at least in Canada and the US, which is what I know as
an American who is a Canadian resident) for each road segment:
1. The street name.
2. The street "direction" (N, S, W, E, NE, SW, etc.) for streets like
West 10th Ave or Deer Trail SE. This value is missing for streets
that do not have a "direction".
3. Address range information that consists of (a) the starting number
on the left-side of the street, (b) the starting number on the right-
side of the street, (c) the ending number on the left-side of the
street, and (d) the ending number on the right-side of the street.
These are often called Left and Right From-To pairs.
4. Local geographic area identifiers. In US TIGER/Line data this is
the five digit zip code. In the UK the likely equivalent would be the
first block of characters (e.g., NN1, or AB32) of the postcode. In
Canada the ideal combination is the first three characters of the
postal code _and_ the "city". Both are needed in Canada for handling
sparsely populated rural areas, which we have a lot more of than is
the case in the UK ;-). The local geographic identifiers can be
different for one side of the street versus the other. The dividing
line between Vancouver and Burnaby, British Columbia is a street
called Boundary Road. In this case, both the city and first three
characters of the postal code depend on whether you are talking about
a location on the west or east side of the street.
With this set of information one minimizes privacy concerns since the
address information is very incomplete, but there is enough
information to do things like geocode a particular street address
with a reasonably high degree of accuracy (not sub-metre, but not
bad). Other bits of information that would be nice additions to each
road segment are:
1. Whether the street is one-way, and the direction of travel on that
street if it is one way.
2. The posted speed limit on the road.
3. The number of lanes of traffic.
4. Whether the road has a bike lane.
5. Whether the road has a sidewalk.
From this additional information the road map can then be used for
routing/road directions (customized for walkers, bikers, or drivers)
and constructing drive-time, walk-time, bike-time polygons.
How would I collect this data? For postal codes I would do exactly as
proposed. In Canada and the US people don't generally mind telling
you their postal/zip codes. I would include an explanation of why you
are doing this, and an indicate to people that the data will treated
in a way that will protect their privacy, indicating the way that
this will be done. One thing you may run into (based on what I know
from CA/US) is that in urban areas a particular block may have
multiple postal codes. I've already mentioned what can happen on
different sides of a street with respect to postal codes. It turns
out that heavy mail recipients may have their own postcode that is
different from other locations in the same block, on the same side of
the street. For instance, in residential areas with mixed single
family houses and apartment buildings, an apartment building may have
one postal code, but the single family houses on either side of it
may have a different postal code that is common to both houses. In
terms of the basic road segments, that is what OSM already does so
well. The major change I would make is increasing the emphasis on
street intersections. I would specifically geocode the location
(possibly from all corners) of an intersection and then note the
street name, street direction, street number, the one-way road
information, the sidewalk/bike lane information at each corner. It
seems natural to me to define a road segment as running from one
intersection to the next intersection. If it is done this way, then
it will be pretty easy to attach the left and right from-to pairs of
street numbers, street names, street directions, and one-way road
indicators to the road segments. It turns out that the intersection
data is valuable in-and-of-itself since it allows for easy geocoding
of intersections as opposed to addresses. I might know that a store
is near the corner of 4th Ave and Main St, but don't have a clue as
to its exact address.
While the above is based on my CA/US experience, my guess is that
nearly all of it will be applicable for the UK as well (I'm not sure
if the street "direction" information is applicable). Having read
over some of the Royal Post's documents, there is a lot of similarity
with how things are done in all three countries. A full Canadian
postal code, UK postal code, and US zip+4 code get down to the block-
face level (e.g., one block on one side of the street) in most cases
in all three countries.
Sorry for the length of this post.
On 30-Mar-06, at 1:55 AM, Nick Hill wrote:
> Immanuel.scholz at gmx.de wrote:
>> I don't know whether all of these people agree to have their
>> address listed in an available online database (regardless of the
>> fact that their names are removed, which could be easily obtained
>> from other sources).
> Ultimately, if we are building an on-line database with road names,
> locality names, postcodes and street numbering, all addresses will
> appear on it.
> If we accept this, then whether we use an address to help structure
> the system perhaps becomes practically neutral.
> We then have the point of law Copyright: Does a singular address
> provided to someone constitute a copyrightable work? If so, when I
> have that address, who owns the copyright?
> talk mailing list
> talk at openstreetmap.org
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the talk