On 14 January 2016 at 22:43, Rob Nickerson wrote:
> 2. Matching algorithm updated
> It looks like Robert W updated the matching algorithm shortly afterwards as
> all percentages have seen an increase. London WC is now showing as 100%
> complete but the detailed page gives another story. Don't forget to use the
> ref:edubase (and similar) tags as Robert's tool makes use of these.
Yes, I've made some changes to the tool at
http://robert.mathmos.net/osm/schools/ , but it's to the data imported
from OSM, rather than the (still very simplistic) matching algorithm.
I'm now including both relations and ways tagged with either
amenity=school or amenity=college. (Previously it was just ways tagged
with amenity=school.)
Looking at some of the Edubase data near to me, it's clear that some
of the institutions should be tagged as colleges in OSM. Also near me
are a number of schools in OSM that were tagged using multipolygon
relations to collect together individual ways that make up the
boundary. I've deliberately excluded nodes to encourage these to be
converted into areas.
The matching algorithm could certainly be improved significantly, and
hopefully I'll get a chance to look at this at some point later this
month. But in the mean time, the best way to ensure the right things
are matched to each other is to map any missing schools, retag any
objects incorrectly tagged as schools, and add ref:edubase,
ref:seedcode and ref:deniirn tags as appropriate.
Finally, if you haven't looked recently, you'll find some other other
reports and Overpass Turbo links at
http://robert.mathmos.net/osm/schools/progress/ , which people may
find useful for more Quality Control work on our Schools data.
Robert.
--
Robert Whittaker
___
Talk-GB mailing list
Talk-GB@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-gb
