16.12.2017 03:00, Frederik Ramm пишет:
Hi,
On 12/15/2017 11:31 PM, Damian Rollison wrote:
We ask that you please take a look at the data at the following site and
check the tagging and locations:
I have little knowledge of Walmart and therefore should leave the floor
to people with local knowledge. I see that another commenter has
already assessed your contribution to be superior with no questions
asked within an amazingly short time. I hope you will find more diligent
reviewers like him; careful reviews like that are what make an import a
success for both businesses and the community.
To be clear, the only person from OpenStreetMap involved with the import
until yesterday was me. I consulted Brandify on using osm_conflate, and
most of the work they did themselves. I'd say this is much better than
any of the previous attempts at importing businesses in OSM.
1. Do you happen to have building outlines too, or is there a process by
which whoever adds the thing to OSM can quickly add the building outline
from aerial imagery? It seems like not too much extra work, if one looks
at 4000 POIs anyway, to quickly trace a building to go with it.
OSM Conflator works only with nodes, except when it updates tags on
polygons. If you like mapping building outlines, please do a challenge
for MapRoulette after this import. Brandify doesn't have building outlines.
2. In my particular example, there seems to be a pharmacy in OSM that is
mapped separately, but also called "Wal-Mart". The import does nothing
to the pharmacy. I don't know if all Wal-Marts have a pharmacy - the
pharmacy POI is certainly within the Wal-Mart building. Is that a
completely separate issue, or would e.g. your opening times information
apply to the pharmacy as well?
Some Walmarts have pharmacies and other similar branded sub-stores. I
think Brandify mentioned that data when we talked. But they obviously
decided to start with more important POIs. We might see an import for
these pharmacies later.
3. Personally I am a bit skeptical about the shop names; I see that in
my example you are changing "Wal-Mart" to "Walmart Supercenter" and I
assume that Wal-Mart will, like many other chains, have different
classes of supermarkets and will have different trade names for these -
super, hyper, metro, local, whatever.
There are two types of Walmarts: "supercenters" and "neighbourhood
markets". Often these names are printed on the side in full, see:
http://www.ktrs.com/wp-content/uploads/2016/05/WALMART.jpg
In OSM, one third of already mapped walmarts (1088) are named "Walmart
Supercenter". That's why I recommended to keep the full name instead of
dropping the second word. Also, there are regular "Walmart"s in the
import, so maybe it's best to differentiate between the two.
4. In my particular example, you have set an "addr:full" tag that
contains the street and house number, but you have not set
"addr:housenumber" or "addr:street". What we would normally do is have
"addr:housenumber" and "addr:street" (where we would neither abbreviate
the "East" nor the "Road" in "East Franklin Road"), and the "addr:full"
is optional. If "addr:full" is used (which is completely optional), it
would normally contain city and post code as well.
De-abbreviating street names and distinguishing "Fourth Street of Oaks"
from "4, Street of Oaks" is a very complex task, and unrelated to the
importing subject. I thought the street address from the source data
would be better added to some tag on an object, but without touching
addr:street and addr:housenumber. And addr:full looks good for this.
There is no rules for this tag, and city and country are usually
geocoded, so the value can help when searching.
Denis, thanks for the suggestions. Adding the "operator=Walmart" tag is
a great idea. Not sure about parsing addresses though.
Georg and Damjan, thanks for the phone number suggestion. Adding "+1-"
to these tags shouldn't be hard.
Not sure about the website pattern. I like the "/details" better, but
whether that could be changed for the import, depends on Brandify.
Ilya
_______________________________________________
Imports mailing list
[email protected]
https://lists.openstreetmap.org/listinfo/imports
