[Imports] Ethiopia POI Import
Alex (AddisMap.com)
alex at addismap.com
Sun May 8 11:32:04 UTC 2022
Hi Claire, Hi Imre,
Thanks for your responses.
Claire Halleux wrote:
> Regarding crowdsourced importing tools, HOTOSM tasking manager coupled to the OSMSharp service developed by Ben Abelshausen (and recently added to HOTOSM) works well for reviewing multiple objects within a bbox. It easily loads the objects for review in a separate layer. Alternatively, for checking one object at a time, MapRoulette is a great option.
I searched for OSMSharp and found only a low-level C# implementation
for OSM. Where can I check how this works coupled to HOTOSM tasking
manager? Would this work on smartphones? Because at the end of the
day, the POIs need to be verified on the ground. In the best case,
this would be as simple as on StreetComplete.
Thanks Imre for the thorough analysis.
Imre Samu wrote:
> based on https://en.wikipedia.org/wiki/Subdivisions_of_Ethiopia
> - "woredas" (districts) ---> addr:district ?
> - "kebele" (wards) ---> addr:ward ?
Yes it might absolutely make sense to map those.
> 2.) I have found strange house numbers; Is it valid numbers?
> $ osmium tags-count AddisMapPOI.osm '*housenumber=*' | head
> 3845 "addr:housenumber" "New"
Most of those tags can probably be cleared. But there are actually
houses with the number sign "New", when a house number was not yet
assigned.
> 3.) I have found some strange POI names.
> $ osmium tags-count AddisMapPOI.osm '*=*losed*' | head -n 30
> 143 "name" "Closed Sook"
probably this should be converted to a disused:shop:kiosk tag and
would also need manual verification.
> $ osmium tags-count AddisMapPOI.osm '*=+251*' | head
> 3441 "contact:phone" "+251 "
> 20 "contact:phone" "+251 AV"
> 10 "contact:phone" "+251 NA"
We can safely remove such tags ...
> 7.) The elimination of duplicates - so harder ;
> for example - this will be imported:
Probably this needs lots of manual work, as due to the transliteration
there are often different spellings. Wondering if there are scripts
which use a fuzzy search for identifying such cases and make the
manual duplicate check more efficient.
Regards,
Alexander
Am So., 8. Mai 2022 um 02:13 Uhr schrieb Imre Samu <pella.samu at gmail.com>:
>
> Hi Alexander!
>
> Thank you for opening the data! :-)
>
> Quick feedback about AddisMapPOI.osm.bz2
> - Probably some issues/comments are related.
> - need some data cleaning / re-tagging ..
> - and de-duplicates is a key!
>
> my suggestion:
> - split/partition the data by POI type ..
> - and let's first focus on the more stable POI types
> -- Diplomatic, Banks, College, ...
>
>
> details:
> 1.) Addresses: is it possible to clean the tags? most of them are not documented in the osm wiki; so not usable for most of the programs.
>
> $ osmium tags-count AddisMapPOI.osm | grep addr
> 33260 "addr:subcity"
> 33088 "addr:woreda"
> 29356 "addr:description"
> 24033 "addr:housenumber"
> 21120 "addr:street"
> 176 "addr:city"
> 11 "addr:postcode"
> 10 "addr:kebele"
> 7 "addr:Kebele"
> 7 "addr:housename"
> 5 "addr:buildingname"
> 4 "addr:p.o.box"
> 2 "addr:floor"
> 2 "addr:full"
> 2 "addr:wereda"
> 1 "addr:country"
> 1 "addr:mobilephone"
>
> based on https://en.wikipedia.org/wiki/Subdivisions_of_Ethiopia
> - "woredas" (districts) ---> addr:district ?
> - "kebele" (wards) ---> addr:ward ?
>
>
> 2.) I have found strange house numbers; Is it valid numbers?
>
> $ osmium tags-count AddisMapPOI.osm '*housenumber=*' | head
> 3845 "addr:housenumber" "New"
> 655 "addr:housenumber" "new"
> 531 "addr:housenumber" "NA"
> 451 "addr:housenumber" "AV"
> 141 "addr:housenumber" "1"
> 133 "addr:housenumber" "shed"
> 108 "addr:housenumber" "2"
> 95 "addr:housenumber" "Shed"
> 91 "addr:housenumber" "4"
> 79 "addr:housenumber" "3"
>
> 3.) I have found some strange POI names.
>
> $ osmium tags-count AddisMapPOI.osm '*=*losed*' | head -n 30
> 143 "name" "Closed Sook"
> 72 "owner" "Closed"
> 22 "name" "Sook (closed)"
> 20 "name" "Sook(closed)"
> 19 "owner" "Closed Sook"
> 13 "name" "Closed"
> 12 "name" "(closed)Sook"
> 9 "name" "(Closed)Sook"
> 7 "name" "Sook (Closed)"
> 6 "name" "Sook(Closed)"
> 5 "owner" "No Information/closed"
> 4 "name" "Closed Café"
> 4 "name" "Sook Closed"
> 3 "name" "Closed Bar"
> 3 "name" "Closed Grocery"
> 3 "name" "Closed SOOK"
> 3 "owner" "(closed)"
> 2 "name" "Abdela Sook(closed)"
> 2 "name" "Abeba Sook Closed"
> 2 "name" "Closed Chat Bet"
> 2 "name" "Closed Wholesaler"
> 2 "name" "Kurs Bet Closed"
> 2 "name" "Pepsi Sook(closed)"
> 2 "owner" "(Closed)"
> 2 "owner" "Closed Café"
> 1 "contact:phone" "+251 Closed"
> 1 "name" "(Closed Sook)"
> 1 "name" "(Closed)Grocery"
> 1 "name" "(closed Chatbet"
> 1 "name" "(closed) Grocery"
>
>
> 4.) I have found some strange phone numbers:
>
> $ osmium tags-count AddisMapPOI.osm '*=+251*' | head
> 3441 "contact:phone" "+251 "
> 20 "contact:phone" "+251 AV"
> 10 "contact:phone" "+251 NA"
> 4 "contact:phone" "+251 911"
> 4 "contact:phone" "+251 N.A"
> 3 "contact:phone" "+251 115510694"
> 3 "contact:phone" "+251 115529979"
> 3 "contact:phone" "+251 11652331"
> 3 "contact:phone" "+251 116632828"
> 3 "contact:phone" "+251 9"
>
> 5.) osmium tags-count AddisMapPOI.osm 'amenity=embassy'
> 20 "amenity" "embassy"
>
> As I know the "amenity=embassy" is deprecated, so need to convert the new tagging
> see: https://wiki.openstreetmap.org/wiki/Tag%3Aamenity%3Dembassy
>
>
> 6.) Now the source_ref link is "POI not found" ( ~ 404 ); in this case no need to import.
>
> cat AddisMapPOI.osm | grep http | head -n2
> <tag k='source_ref' v='http://pris.map.et/2014-12-07-133039/pic00787.jpg' />
> <tag k='source_ref' v='http://pris.map.et/2014-12-07-133039/pic00264.jpg' />
>
>
>
> 7.) The elimination of duplicates - so harder ;
> for example - this will be imported:
>
> $ osmium getid --no-progress -f osm AddisMapPOI.osm n228824
> <?xml version='1.0' encoding='UTF-8'?>
> <osm version="0.6" upload="false" generator="osmium/1.14.0">
> <node id="-228824" lat="9.0469216" lon="38.7619826">
> <tag k="FIXME" v="Removed ???"/>
> <tag k="name" v="Karl-Marx-Monument"/>
> <tag k="name:de" v="Karl-Marx-Denkmal"/>
> <tag k="tourism" v="attraction"/>
> </node>
> </osm>
>
> but this node already exists: https://www.openstreetmap.org/node/2721720441
>
> <osm version="0.6" generator="CGImap 0.8.6 (2724028 spike-08.openstreetmap.org)" copyright="OpenStreetMap and contributors" attribution="http://www.openstreetmap.org/copyright" license="http://opendatacommons.org/licenses/odbl/1-0/">
> <node id="2721720441" visible="true" version="2"
> changeset="21144484" timestamp="2014-03-16T20:40:10Z" user="Krteček" uid="272351" lat="9.0469216" lon="38.7619826">
> <tag k="historic" v="memorial"/>
> <tag k="memorial:type" v="statue"/>
> <tag k="name" v="Karl-Marx-Monument"/>
> <tag k="name:de" v="Karl-Marx-Denkmal"/>
> </node>
> </osm>
>
> 8.) duplicated hotel example:
>
> import candidate: Foyat Hotel
>
> $ osmium getid --no-progress -f osm AddisMapPOI.osm n302338
>
> <?xml version='1.0' encoding='UTF-8'?>
> <osm version="0.6" upload="false" generator="osmium/1.14.0">
> <node id="-302338" lat="8.991262" lon="38.794178">
> <tag k="contact:email" v="info at foyathotel.com"/>
> <tag k="mobile" v="+251 966 21 54 32, +251 966 21 54 33, +251 911 52 51 74"/>
> <tag k="name" v="Foyat Hotel"/>
> <tag k="phone" v="+251 11 660 70 96"/>
> <tag k="tourism" v="hotel"/>
> </node>
> </osm>
>
> in the OSM database already duplicated (2x ) :
> https://www.openstreetmap.org/node/4317313595
> lat="8.9912752" lon="38.7942496" "name="FOYAT HOTEL"
> https://www.openstreetmap.org/node/4298833201
> lat="8.9913672" lon="38.7942190" "name="Foyat Hotel"
>
> Regards,
> Imre
>
> Alex (AddisMap.com) <alex at addismap.com> ezt írta (időpont: 2022. máj. 5., Cs, 20:33):
>>
>> Hi,
>>
>>>
>>> For start, publish this dataset to allow judge its quality.
>>
>>
>> sure, http://download.addismap.com/poi/AddisMapPOI.osm.bz2
>>
>>>
>>> I am highly dubious that adding cafes, kiosks and hotels
>>> based on up to 12 year old data is a good idea.
>>>
>>> I am unfamiliar with Ethiopia, but in Europe in such old data
>>> very large part of object would be not existing anymore.
>>
>>
>> The question is if the benefit from having that data is higher than not having it :-) We are thinking about importing the data in a way it does not hurt but can be easily updated with existing OSM apps (for example StreetComplete).
>>
>>>
>>> Also, you would need to contact local community and ask
>>> whether importing this dataset is welcome
>>> (I see that you posted to mailing list, see
>>> https://wiki.openstreetmap.org/wiki/Contact_channels
>>> for listings of other possible communities).
>>
>>
>> Yes, we have some contacts here and will reach out on different channels.
>>
>>> I would be opposed to this import, but if local community
>>> disagrees and wants to add this data - feel free to ignore this.
>>
>> okay
>>
>>>
>>> Note that imports would require elimination of all duplicates.
>>
>>
>> okay, sure, yet we are wondering what's an efficient workflow for that task. Advice is appreciated :-)
>>
>>> > The check_date would be set to 2016-01-01
>>>
>>> It should be set to date when object was confirmed to exist.
>>
>>
>> We don't have the exact date here, just a range. We might be able to reconstruct the real check date from other datasets, or maybe can set it to 2010 to be on the safe side.
>>
>> Regards,
>> Alexander
>>
>>
>>
>> _______________________________________________
>> Imports mailing list
>> Imports at openstreetmap.org
>> https://lists.openstreetmap.org/listinfo/imports
More information about the Imports
mailing list