[Talk-GB] Quarterly Project: Addresses and Postcodes

Lester Caine lester at lsces.co.uk
Thu Oct 19 15:32:03 UTC 2017

On 19/10/17 15:37, Colin Smale wrote:
> Which boundaries are your referring to, which have yet to be mapped?
> There are big holes in Civil Parish + Community mapping in the north of
> England/Wales/Scotland, but most of England is OK. AFAIK all other admin
> boundaries are in there.
Parish boundaries are the one that normally catch me out ...
And often it's difficult to sort ward boundaries from one another.

> "Place" boundaries are a whole other can of worms, because they have no
> defined boundaries in most cases and most of the UK will be in the ether
> between places. They will usually differ from Royal Mail´s perspective
> anyway.
Hospital grounds and university campus boundaries are another area that
are improving, along with industrial estates, but one I look after does
not fair well with Nominatim as its WR11 post codes inside Gloucestershire.

> If the use case is to get decent results out of nominatim, we need to
> have a discussion about how best to approach that. I think we will not
> get there with OSM data alone - changes to nominatim's logic will be
> required. But it all depends on your expectations I suppose.

'Places' like Wychavon being used for directions are simply wrong, and
more of a problem often is finding places on OSMAND that are currently
not showing up properly. I had a run up to Haydock .. Wedge Avenue ...
but it does not appear. Adding postcodes around there will probably
help, but having selected 'Haydock' one would expect all the roads to be
listed? Not sure how OSMAND is holding data, but a search on
fails ... this is not really even a 'post' problem, just finding directions.

> On 2017-10-19 16:27, Lester Caine wrote:
>> On 19/10/17 14:31, Dave F wrote:
>>> On 19/10/2017 12:04, Lester Caine wrote:
>>>> On 19/10/17 11:35, Adam Snape wrote:
>>>>> Doesn't its location within the UK make an explicit UK tag unnecessary?
>>>> But when reading a single object tags do you know just where it is? Some
>>>> other mechanism has to return the 'inside boundary' data which takes
>>>> processing power.
>>> OSM is geospatially aware. Nominatim have stated that it's not intensive
>>> processing & prefer it over is_in*. If unwilling to use 'inside
>>> boundary' coding it requires *every* object to have multiple location
>>> tags for *every* search boundary. Expecting mappers to add this enormous
>>> amount of data is selfish.
>> The reverse of that is that there are a large number of boundaries that
>> have yet to be mapped and in some instances may be difficult to map at
>> all. I'm not suggesting that mappers add any more tags than useful and
>> easy to add. What I AM suggesting is that 'Expecting mappers to add this
>> enormous amount of data is ...' unnecessary when some key tags will
>> cross reference the rest of the data! And where linked data changes then
>> one does not have to address every object, just the top record.
>> Yes automation can manage and change multiple records in parallel and
>> fill multiple tags from the one entry, but does all that information
>> have to be stored raw in OSM WHEN the processing can just as easily
>> provide it?

Lester Caine - G8HFL
Contact - http://lsces.co.uk/wiki/?page=contact
L.S.Caine Electronic Services - http://lsces.co.uk
EnquirySolve - http://enquirysolve.com/
Model Engineers Digital Workshop - http://medw.co.uk
Rainbow Digital Media - http://rainbowdigitalmedia.co.uk

More information about the Talk-GB mailing list