[Talk-GB] RFC Mechanical edit: shop=betting to shop=bookmaker for selected names

SK53 sk53.osm at gmail.com
Thu Oct 23 10:16:23 UTC 2014

I have no particular objection to this mechanical edit. I would query
whether it is that useful.

When I write code for consuming retail data I have no problems at all in
adding additional simple rules for assigning shops to particular classes. A
basic starting point for such work is looking at the available list of
values from taginfo or the extract I am using. It is quite usual to have
clauses such as CASE WHEN shop in ('bookmaker','betting') THEN ..., and it
is similarly easy to perform similar tag manipulation in tools such as LUA
(for osm2pgsql pre-processing) or mkgmap for creating Garmin maps.

The substantial problem is not synonymy of widely used tags, but a long
tail of other tag usage. Simply querying for places with node=Coral using
Overpass I immediately (just by clicking on a couple of nodes) found a
shop=bookies tag. Taginfo shows a range of values
<http://taginfo.openstreetmap.org/search?q=betting#values> which form a
long tail of other much less widely used synonyms. Doing a full query on
all elements with one of the names proposed gives the following result:

  bookmaker William Hill
175  bookmaker Ladbrokes
137  bookmaker Coral
107  betting William Hill
79  betting Ladbrokes
68  betting Coral
William Hill
28  bookmaker Paddy Power
13  bookmaker William Hill gambling 11  betting Paddy Power
11  gambling William Hill
9  bookmaker Ladbrokes gambling 7  bet Coral
Ladbrokes gambling 6  gambling Ladbrokes
6  betting Ladbrokes gambling 5  betting William Hill betting 5
Coral gambling 5  bet Ladbrokes
William Hill gambling 5  betting BetFred
5  gambling Coral
4  bookmaker BetFred
4  yes Ladbrokes
Paddy Power
3  travel_agency Coral
2  betting Coral betting 2  betting Coral betting_shop 2  bookmaker Coral
gambling 2  bookmakers Ladbrokes
Coral casino 2  bet William Hill
2  bet Paddy Power
2  bookmakers Coral
Coral bookmaker 2  yes William Hill bookmaker 1  turf_accountant William
Paddy Power bookmakers 1  kiosk Ladbrokes
1  pawnbroker Coral
1  yes Coral
1  betting BetFred betting 1  betting Ladbrokes betting 1  betting Ladbrokes
betting_shop 1  bookmakers William Hill
William Hill casino 1  beting William Hill
1  hardware Ladbrokes
1  yes Paddy Power
1  bookmaker Ladbrokes betting shop 1  gamabling William Hill
1  bookies Coral
1  video_games Ladbrokes
1  general William Hill
1  betting William Hill gambling 1  betting_shop William Hill
1  betting William Hill betting_shop 1
Ladbrokes casino 1  betting_shop Coral
Ladbrokes bookmakers 1  alcohol Ladbrokes
1  Betting William Hill gambling 1  bookmarkers William Hill gambling 1
bookmaker Coral betting 1  gambaling Coral

Apart from typos, use of the amenity tag is quite common, as is no
functional tag at all just a name. Excluding those combinations covered by
shop = betting or shop = bookmaker this still leaves another 25 possible
tag values combinations which are highly likely to be associated with a
bookmaker. These actually represent about 1/6 of all such elements in Great

Fixing these values would be more valuable than making a correction which
is readily handled in data consumer code. However, much of this fixing
cannot be casually done with a mechanical edit, particularly when an
element has not received any functional tags or when the tag might be one
of confusion, but equally might be correct.

I will provide a link to a shape file of centroids of the 'funny' values.

In summary, tidying up synonymies like this really does not make life much
easier for data consumers. The harder work is in identifying: less widely
used synonymies, perhaps things where the tag had not been established , or
places which were marked for later enhancement which was not carried out,
etc. Working to resolve these would be much more valuable as this is not
something a data consumer can resolve with a few characters of code unless
they have a very specific interest in that class of data.



PS. There may be something in looking at particular classes of retail on a
regular basis and identifying the 'funnies' for further checking. This
could be done by OSMers in Great Britain as a whole, and like many such
activities would probably result in other improvements on the side.

On 22 October 2014 23:04, Matthijs Melissen <info at matthijsmelissen.nl>

> Dear all,
> For all objects tagged with shop=betting and name Betfred, Coral,
> Ladbrokes, Paddy Power or William Hill, I am planning to change the
> tag shop=betting into shop=bookmaker.
> Please see
> https://wiki.openstreetmap.org/wiki/Mechanical_Edits/Math1985/Betting
> for more information.
> Please let me know if you have any comments. If there are no further
> comments, I will invite list members to vote on this automatic edit. I
> will not proceed without at least 8 votes with 2/3 approval.
> Kind regards,
> Matthijs
> _______________________________________________
> Talk-GB mailing list
> Talk-GB at openstreetmap.org
> https://lists.openstreetmap.org/listinfo/talk-gb
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk-gb/attachments/20141023/65dc391f/attachment-0001.html>

More information about the Talk-GB mailing list