[Talk-GB] OSM Analysis Updated with latest OS Locator data
danstowell+osm at gmail.com
Wed Nov 19 19:40:29 UTC 2014
2014-11-19 19:14 GMT+00:00 Dave F. <davefox at madasafish.com>:
> On 19/11/2014 09:42, Shaun McDonald wrote:
>> There have been many places knocked off the 100% completeness, with 17
>> areas now below 95% complete. Overall completeness is currently 98.00%.
> Thanks for that. It's a really useful resource to check against. However...
> No gazette/database is ever fully accurate. Everything contains errors. So
> I'm somewhat surprised when I see an area listed as 100%. It implies users
> have blindly transferred across without checking in the real world. For
> example in my area there's a street supposedly called 'roman road' I'm born
> & bred & never heard it called that. There's no roads or building signs, the
> local authority, emergency services, or newspaper ever describe it that way.
> A couple of others are where named rows of buildings are assumed to be the
> street name, but are signed differently on the ground.
> Dave F.
Where OS data doesn't match OSM properly-surveyed data, we can put
OS's version in "not:name" if it's just wrong, or "alt_name" if it's
known but not the main name. Shaun's analysis can make use of these
other tags to avoid counting differences that have already been
checked. If you look at one chosen area you can see a separate column
for "not name" in Shaun's analysis - so it looks to me like they are
indeed using my not:name annotations, hoorah! Look at all these
entries which are NOT causing false positives!
More information about the Talk-GB