[Imports] Import suggestion/proposal: street address data for the City of Vancouver, British Columbia (Canada)

[Adam Williamson]

Hi there, OSM folks!

I'm a fairly new OSM contributor residing in Vancouver (BC, Canada). I'd
like to suggest/propose the import of the City of Vancouver's street
address data set to OSM. I have created an import page on the wiki at
http://wiki.openstreetmap.org/wiki/User:AdamWill/Imports/Vancouver_property_addresses containing all the details.

Briefly, the data is available from the City at
http://vancouver.ca/your-government/open-data-catalogue.aspx , under a
license which can be classified as 'OK for OSM use but requires a
specific statement from the City', as I understand it. More details on
the wiki page.

The raw data consists of just north of 100,000 single nodes - mainly
placed in the centre of property lots, I believe - containing the street
name and house number. The quality of the data is described by the city
as "Much of the City's land base is created using survey accuracy
however some features are not as precise." - this text doesn't cover
only the street address data, note, so it may be that *all* the street
address data is 'survey accurate' or (less likely) *none* of it is. I
can ask the CoV's open data team for clarification on the claimed
accuracy of the street address data specifically, if desired.

There is no complete or reliable building shape data available for the
CoV from any source of which I'm aware: the City provides a small amount
of shape data for the downtown peninsula, but hedges it around with
disclaimers (it's outdated, incomplete, and not entirely reliable,
apparently), so for all intents and purposes all we have for building
shapes is manual survey. (Looking through other address import proposals
this seems to be something that comes up, so I thought I'd throw it in).
There is, however, high-quality free satellite data available for the
area, which helps a lot with checking of the data.

I have written and provided on the wiki page the necessary ogr2osm
translation file and a small bash script to convert the dataset to a
format appropriate for OSM and clean it up: correct tag names and drop
some incomplete and duplicated entries. The translation file and script
are freely licensed and heavily commented. Full details on the wiki
page.

There is some existing data in the CoV, but not a lot, and this dataset
would be a huge advance: after cleaning it contains ~93,000 addresses,
while according to overpass-turbo, we currently have <1,000 objects with
addr:housenumber and <2,000 with addr:street. Those numbers are small
enough that manual (but tool-assisted, of course!) conflation resolution
ought to be relatively simple, I believe.

I'd more than welcome any assistance and advice on this project! Thanks
very much :)
-- 
adamw