[Talk-ca] duplicate address data

Thu Mar 26 19:41:20 UTC 2015

Hi John,

thanks for the quick feedback.

I can try that by adding some rules to mkgmap so that it
ignores data with source=CANVEC* that is not like CANVEC 10 .
If I got you right I should only find a a few error messages.
If that turns out to be correct, I'll try to learn more about the imports.

Ciao,
Gerd

Date: Thu, 26 Mar 2015 15:20:36 -0400
Subject: Re: [Talk-ca] duplicate address data
From: jwhelan0112 at gmail.com
To: gpetermann_muenchen at hotmail.com
CC: talk-ca at openstreetmap.org

Basically the CANVEC data imports need cleaning up.  I think there were ten different versions, the most common I think is 7.  Unfortunately some mappers locally removed the CANVEC tags from the data if they touched it or even on the import as they didn't think it was important.  Sometimes addresses were imported with a CANVEC tag, sometimes not.  Due to funding cutbacks the CANVEC data is no longer exported in OSM format.

Also there was some original mapping done from low resolution satellites, Yahoo I think provided the images so some roads were mapped about 100 meters from where they should be, where highways had been mapped the CANVEC imports were sometimes used and sometimes not.  In Ottawa we took a local decision to delete all the roads above service roads and replace them with CANVEC imports because of the data quality issues of the existing road network and that was some years ago.

Unfortunately in Canada we have fewer mappers per kilometer of highway than in Germany and the CANVEC imports were very useful.

The clean up solution I would suggest would be to delete all address information with a CANVEC tag on it then import only CANVEC 10 which is the latest version but that's a lot of work but the end result would be clean.  It might also hit problems as the address information would follow the CANVEC highways rather than those highways mapped in other ways but it would only be the address information and the road network would remain as it is.

Cheerio John

On 26 March 2015 at 14:54, Gerd Petermann <gpetermann_muenchen at hotmail.com> wrote:

Hi list,

I am one of the developers of mkgmap, see also
http://wiki.openstreetmap.org/wiki/Mkgmap
and
http://gis.19327.n5.nabble.com/Mkgmap-Development-f5324443.html

During the last weeks I've enhanced the support for 
the evaluation of addr:interpolation ways or more general
the evaluation of addr:housenumber, addr:street and so on
to improve the address search in Garmin devices for maps
based on OSM data.

I live in Germany and I am not very familiar with the address
schemes used in North America.

It turned out that data in Canada is very special because of the 
CanVec imports. I find a huge amount of addr:interpolation 
ways that seem to make no sense, often those are 
duplicated with identical or nearly identical points 

Example: The ways 
http://www.openstreetmap.org/way/99649911
and
http://www.openstreetmap.org/way/83504524

One has source=NRCan-CanVec-7.0, the other source=CanVec 6.0 - NRCan
Is there a good reason for this redundancy?
If not, what is the best way to remove these duplicates?
I can think of different ways:
1) keep only the eldest entry
2) keep only the youngest entry
3) keep the older and add a note that the data is confirmed by NRCan-CanVec-7.0 

Second problem that occurs very often:
http://gis.19327.n5.nabble.com/weired-housenumbers-in-Canada-tt5835196.html

The example still exists:
http://www.openstreetmap.org/way/133338396
A long addr:interpolation way connecting two points which both have 
addr:housenumber=5.
If that data is correct, what information does it offer?
I can only guess that along this long way one can find a house with
number 5. Or does it mean that the house is in the middle?
Or is the whole ground along this road "20th Sideroad 5" ?

And what does it mean when multiple addr:interpolation ways
exist connecting points with equal addr:housenumber,
like here:
http://www.openstreetmap.org/way/133570749#map=18/45.68026/-80.03331&layers=N
Where is  Bear Hug Lane 10 and why are there so many addr:interpolations ways for it?

Any help is welcome. 

Gerd

P.S.
The program mkgmap in the housenumber2 branch creates a log file that reports the problem cases in a format like this:
"found additional addr:interpolation way with same meaning, is ignored: 1st Avenue http://www.openstreetmap.org/way/99649911 127..123, step=2"
it would be easy to change that if needed.

_______________________________________________

Talk-ca mailing list

Talk-ca at openstreetmap.org

https://lists.openstreetmap.org/listinfo/talk-ca

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk-ca/attachments/20150326/413ae1ac/attachment-0001.html>