[Talk-GB] Errors in Street Names in Addresses

[Robert Whittaker (OSM lists)]

On 27 January 2018 at 20:09, Robert Whittaker (OSM lists)
<robert.whittaker+osm at gmail.com> wrote:
> Following on from last quarter's post-code and addresses project, I've
> been doing a bit more playing around with UK address data from OSM.
> I've put together a new report at
> http://robert.mathmos.net/osm/addresses/street-warnings/ which
> provides lists of potential issues in street names that appear in
> addresses on OSM objects.

Following some of the comments here and after looking at more of the
data, I've made a couple of tweaks to the tool:

1/ Relations with type=street are now supported in the same was as
type=associatedStreet was before. So a street name given in a 'street'
relation will now be picked up.

2/ addr:postcode tags on highway=* objects should now be listed in a
separate section, and the name=* tag is used to find the street name.
So if you want to ignore such objects you can.

(There weren't nearly as many objects in case 2 as I thought there
would be here based on people's comments, so it's possible I've messed
up the programming logic somewhere. If there are still any objects
with a highway=* tag listed in other sections, then please let me
know, and I'll see if I can fix the bug.)

Finally, there was a request not to treat "&" as an odd character. I
think this won't occur too often in actual names, and if it's included
in an OSM addr:street tag it's as likely as not to be an inappropriate
abbreviation. So I think on balance it's worth keeping. Remember that
the items listed in the tables are only warnings, and not necessarily
errors. There will inevitably be some false positives in the checks.

Robert.

-- 
Robert Whittaker