[Talk-ca] [Fwd: Compliance statement] - Vancouver address information cleared for OSM use

Adam Williamson adamw at happyassassin.net
Mon Feb 3 18:19:19 UTC 2014

Good news, folks: the City of Vancouver emailed Paul and I the attached
statement this morning, which should clear the Property Information
dataset at
http://data.vancouver.ca/datacatalogue/propertyInformation.htm for use
for OSM and other free projects.

They also said they accepted the concerns that had been raised in the
ODDL discussion of the OGL-BC license -
http://web.archiveorange.com/archive/v/EfptdAbNnlONouk3B79m - and seemed
to be interested in addressing them more systematically by modifying the
license itself, so that could also bode well for the future.

Many thanks to the City for being responsive on this!

I've been very busy with work-related stuff lately so haven't had time
to polish up the proposed Vancouver address import much, plus I didn't
get any feedback on it in public :) Paul and I chatted about it in
person, though, and he noted some concerns:

* that address information per se is more useful for geocoding than it
is particularly map-y data, so it might make more sense to import it to
layers like Nominatim than directly into OSM itself

* that the fact the nodes are located in the middle of property lots (we
think) means the nodes for very large lots - like the golf courses, most
obviously, but also potentially industrial locations, for e.g. (yep,
Vancouver still has some) - are likely not in the best places (we
checked a couple of the golf courses, and yeah, the nodes are out in the
middle of the course, not on the clubhouse or the street entrance)

* that it might be better to try and do a Seattle-style approach to try
and also drive community involvement in OSM, rather than just doing it
in a very geeky data-driven way

I also did some eyeballing of the current imported data, and there are
some places where it conflicts with manually-entered data, most
obviously in the West Broadway corridor someone did a manual survey on a
few years back. I'm about 85% sure the manually entered data is
inaccurate in this case - the building outlines are all offset from the
'Mosaic' aerial photography on Paul's server, and I suspect they were
traced from inaccurate satellite imagery prior to the better Mosaic data
being available, and the addresses were entered on top of the traced
building outlines, hence the inaccuracy. But it would be best to go out
and do a field survey to confirm this, of course.

all of these seem like reasonable concerns, and I'll be looking into
them when I have time. (Particularly in the context of my aim to get an
OpenTripPlanner server going - it should be possible to make the address
data available to such a service without it being imported to OSM, I
just need to find some free time to look into it).

I did make a bit of progress that I didn't note in my last emails,
though, last time I worked on it: after my cleanup of the existing
street addresses in the CoV, we can de-duplicate the data (drop existing
addresses from it, so it doesn't duplicate manually-entered nodes)
pretty much perfectly. I'll try and find a bit of time to update the
proposed import wiki page with the details on doing that soon.

Thanks folks!
-------------- next part --------------
An embedded message was scrubbed...
From: "Van Fraassen, Barbara" <Barbara.VanFraassen at vancouver.ca>
Subject: Compliance statement 
Date: Thu, 30 Jan 2014 18:56:29 +0000
Size: 6770
URL: <http://lists.openstreetmap.org/pipermail/talk-ca/attachments/20140203/c3441d6a/attachment.mht>

More information about the Talk-ca mailing list