[Talk-GB] Geospatial Commission to release UPRN/ UPSN identifiers under Open Government Licence

[Mark Goodge]

On 09/04/2020 14:26, Robert Whittaker (OSM lists) wrote:

> I would have said that ref:uprn and ref:usrn are the natural choices
> for use to use. However, I've seen some calls for country codes to be
> added to 3rd-party ref values, so we might consider ref:UK:uprn and
> ref:UK:usrn instead. This isn't explicitly documented in the wiki at
> https://wiki.openstreetmap.org/wiki/Key:ref though the French
> community seems to be using it, as can be seen at
> https://wiki.openstreetmap.org/wiki/France/Liste_des_r%C3%A9f%C3%A9rences_nationales
> , and I think it might make sense.

I agree that adding a country identifier makes sense. One of the key 
attributes of a UPRN is that it is unique within a country. But it may 
not be globally unique if other countries adopt a similar system. And, 
because it's just an integer, unlike a postcode, you can't infer the 
country from the format. But, on the other hand, adding another layer 
makes it more likely that people will tag them wrongly using what they 
think is the right method simply because that's what's most obvious to them.

> I don't see any value in adding NLPG (or it's incorrectly ordered
> variant NPLG). Although the National Land and Property Gazetteer is
> where the UPRN values originate from, if they're being used as core
> identifiers by the government, they're no longer just NLPG values.

I agree with this, too. The NLPG is just a database of UPRNs and other 
data, it isn't the source of them.

> I also don't see any benefit in adding a :1 :2 etc suffix to the key
> in anticipation of multiple values (which seems to have been done in
> several existing UPRN keys). I think this will actually make it harder
> for data-users than having a single key name and separating multiple
> values with semi-colons. (You would suddenly need to search multiple
> different keys to get all possible UPRN-tagged objects.)
> 
> So I'd propose that we use either ref:uprn and ref:usrn, or
> ref:UK:uprn and ref:UK:usrn. What does everyone else think?

I'd be happy with either, so long as it's consistent.

Mark