[Talk-GB] Mechanically setting name tag of Naptan data

[Matthijs Melissen]

On 27 November 2013 00:39, SomeoneElse <lists at mail.atownsend.org.uk> wrote:
> I can't comment on bus stops in Birmingham, but elsewhere in some cases
> (Notts) "name" seems to have been set to naptan:CommonName - see for here:
>
> http://www.openstreetmap.org/browse/node/502389340

Yes, the situation really from city to city. The main rule is that we
should set the name to whatever is written on the bus stop. In
Birmingham, that is Street+CommonName. In Birmingham, CommonName by
itself is not unique, while in Notthingham it seems to be.

> This far removed from the original naptan import, you'd really need to vet
> each uploaded bus stop, in order to try and reconcile:
>
> o duplications (stops that were mapped before the original import), such as
> http://www.openstreetmap.org/browse/node/502389340 .
>
> o "invalid" changes where someone's used or misused a bus stop node for
> something else (for example, new mappers sometimes see a bus stop with a pub
> name and add "amenity=pub" to it), or where someone's deleted a bus stop by
> mistake.
>
> o "valid" changes where someone's moved a naptan-imported stop to a more
> accurate location, or deleted a bus stop because buses no longer stop there.
>
> o Updated information from Naptan to stops verified (or not verified) in
> OSM.

I will not overwrite stops that have already a name tag. I am also not
doing a new import, I am only updating the names of existing stops
with Naptan data. I think that should avoid these problems.

> A "guided merging" approach has been tried in the past, with DfT cycling
> data:
>
> https://wiki.openstreetmap.org/wiki/Potlatch_2_merging_tool
>
> Maybe something like that could work?

I think merging is not really a problem given that I only look at
stops without name tag, so I think to do it with a script would be
better.

-- Matthijs