[HOT] testing:Taginfo for African, Central-American, etc countries ( test: ~100)

Philippe Verdy verdy_p at wanadoo.fr
Wed Mar 21 18:34:45 UTC 2018


The mapping by "continent" is in fact not needed at all (and
controversial). Mapping by country only is sufficient, notably if you use
standard ISO 3166-1 codes.

If you want to have some internal mapping by continent for Taginfo, you can
still apply such private mapping in the Taginfo website (but other Taginfo
sites may have different views about how to group "countries" that are
viewed by others as dependancies).

Such controversy has already occured on the OSM wiki, and in the OSM
database, so much that mapping continent boundaries in OSM has been
compeltely removed, allowing instead each country to have possibly
overlapping boundaries for what they claim individually, and then in OSM we
added pseudo-countries (at level 2) for areas claimed by several countries.

If you intend to have per-country Taginfo instance, be open to this: avoid
defining what is a "continent". Even the UN has problems in defining them
for its statistics agencies, and sveral agencies used different groupings.
I don't think this is needed, and in fact it will just make domain names
for Taginfo instances unnecessarily long. Just keep the ISO 3166-1 country
code, nothing else. We can stiol have other external web pages to present
these instances with relevant grouping for their own needs, it's not
complicate at all to do on just one table or 5/6 webpages for their
groupings and navigational links.

If you want to have per continent statistics, they will have their own
mappiung and own data tags chosen more internationally and less locally,
but I bet that the international Taginfo instance is sufficient for that
(we may want a Taginfo instance for the European Union for its applied
common policies and required open data that member states must provide, but
even there, there will be extra non-EU data because of of cooperation, e.g.
with EFTA, or data for extra-european dependancies, or cooperation with
candidate members; and UK will probably keep large cooperations in many
domains related to Open Data). So even a "EU" instance would cover a bit
more than just Europe or the European Union...

The same remark will apply to the Asia-Pacific region (should it include US
or Canada, or even Chile ?). What each instance will need to over will not
be necessaily exclusive, and each Tagingo instance will have its own
coverage independant of per-country Taginfo instance.




2018-03-21 19:13 GMT+01:00 Imre Samu <pella.samu at gmail.com>:

> >I wonder if the choice (in domain names) of non standard prefix for
> continental areas before the country code really helps making the instances
> per country really accessible: "ca-ni.*"
> >for example should just be "ni.*", or "as-lk.*" should just be "lk.*",
> there's no need to use a continental area prefix, but if you need some
> prefix,
> > it should be the same for all countries (e.g. "cc-ni.* for Nicaragua,
> and "cc-lk" for Sri Lanka), or you could use submains of a common generic
> domain (e.g. "ni.cc.*" and "lk.cc.*").
>
> Now I am using Geofabrik ~ continents ( https://download.geofabrik.de/ )
> + Russia : https://download.geofabrik.de/russia.html
> and _now_ -  I need to separate the countries by inputs("continents")   -
> and it is like  ~ prefix  or a grouping    ( If I have more resources ( RAM
> /DISK)  it is easy to change )
>
> the (reversed) subdomain is a good idea.   ( "ni.ca.*" )
> or  I can use a longer name:      "nicaragua.central-america.taginfo-dev.*"
>      But the "ISO" code more neutral.
>
>
> > The classification of countries per continent or continental is
> problematic when they have parts covering multiple continental areas (e.g.
> France, UK, US, Turkey, Cyprus, Russia, Venezuela):
>
> I have "low resources" so my first decision was to use Geofabrik
> Continents, not perfect, but it is working.
> -  Turkey, Cyprus, :  in https://download.geofabrik.de/europe.html
> -  outside of UK: usually has an iso code.   (  MS = Monserat:
> http://ca-ms.taginfo-dev.opengeodata.hu/ ;  http://ca-vg.taginfo-dev.
> opengeodata.hu/, ... )
> -  US,CA,FR,DE  - better to use States ( ISO3166-2 )
> -  Now for FR, NL, ES, DK  -  I am using  the https://taginfo.
> openstreetmap.org/keys/ISO3166-2
>
> fr-bl Saint Barthélemy ca-fr-bl.taginfo-dev.opengeodata.hu
> Saint-Barthélemy
> fr-mf Saint Martin         ca-fr-mf.taginfo-dev.opengeodata.hu
> Saint-Martin
>
> nl-aw Aruba ca-nl-aw.taginfo-dev.opengeodata.hu Aruba
> nl-cw Curacao ca-nl-cw.taginfo-dev.opengeodata.hu Curaçao
> nl-sx Saint Martin ca-nl-sx.taginfo-dev.opengeodata.hu Sint Maarten
>
> es-cn  Canary Islands af-es-cn.taginfo-dev.opengeodata.hu Canarias
> es-gc   Las Palmas af-es-gc.taginfo-dev.opengeodata.hu Las Palmas
> es-ml  Melilla af-es-ml.taginfo-dev.opengeodata.hu Melilla
> es-tf          Santa Cruz de Tenerife af-es-tf.taginfo-dev.opengeodata.hu Santa
> Cruz de Tenerife
>
> But this is a simple customization.  If somebody need I can generate ~
> 4000 areas based on ISO3166-2
>
> >So to use these instances we would then need to maintain custom mappings
> in other tools
> > (including on the OSM wiki that attempts to map relevant Taginfo
> instances, for now by mapping some per country instances from languages).
>
> maybe we can create an API  ( and query the working taginfo instances , or
> just link the proxy page   )
>
> In my mind, this is a dynamic ...
>
> in an emergency case we can start a Taginfo for a subarea like:
> https://www.openstreetmap.org/relation/1600775#map=7/-5.693/34.409 and
> can reach via :      "tz-23.africa.*"    or a    "q153326.africa.*"   domai
> ,and we can stop 2 moth later.
>
> But the other integration nothing to do:
> - Overpass Turbo search :  98%  should working
> - Taginfo sources ( Wiki, projects, ... )   same as a "big" taginfo
> https://taginfo.openstreetmap.org/sources
>
>
> >How do you intend to support translations and localisation/adaptation
> (e.g. Bidi layout, recommended minimum font size, and required minimum
> line-height per script)
> >on these per-country Taginfo instances, in countries that have multiple
> languages (e.g. Switzerland, Belgium, Morocco or India) ?
>
> Now this is just a Proof of concept.
> If you have suggestions please create an issue:  https://github.com/
> taginfo/dockerized-taginfo
>
> Imre
>
> 2018-03-21 17:11 GMT+01:00 Philippe Verdy <verdy_p at wanadoo.fr>:
>
>> I wonder if the choice (in domain names) of non standard prefix for
>> continental areas before the country code really helps making the instances
>> per country really accessible: "ca-ni.*" for example should just be "ni.*",
>> or "as-lk.*" should just be "lk.*", there's no need to use a continental
>> area prefix, but if you need some prefix, it should be the same for all
>> countries (e.g. "cc-ni.* for Nicaragua, and "cc-lk" for Sri Lanka), or you
>> could use submains of a common generic domain (e.g. "ni.cc.*" and
>> "lk.cc.*").
>>
>> The classification of countries per continent or continental is
>> problematic when they have parts covering multiple continental areas (e.g.
>> France, UK, US, Turkey, Cyprus, Russia, Venezuela): there's no well defined
>> mapping of countries to a single continental area even for those that are
>> encoded in ISO 3166-1 (with also dependant territories having a secondary
>> ISO 3166-1 code and sevral aliases allocated). Even for America this is
>> quite complex: is the Caribbean only part of North America, what to do
>> about Venezuelan dependencies in the Antillas ?). So to use these instances
>> we would then need to maintain custom mappings in other tools (including on
>> the OSM wiki that attempts to map relevant Taginfo instances, for now by
>> mapping some per country instances from languages).
>>
>> How do you intend to support translations and localisation/adaptation
>> (e.g. Bidi layout, recommended minimum font size, and required minimum
>> line-height per script) on these per-country Taginfo instances, in
>> countries that have multiple languages (e.g. Switzerland, Belgium, Morocco
>> or India) ?
>>
>>
>> 2018-03-20 18:30 GMT+01:00 Imre Samu <pella.samu at gmail.com>:
>>
>>> TLDR: I am working on a "dockerization of Taginfo" for every country and
>>> now I can show you a Proof of Concepts of my idea,
>>> and if you have time please test, give a feedback.
>>> And I am searching sponsors/funds/microgrants/*  for hosting/dev   if
>>> this project is useful.
>>> Temporary link ( for the next 2 week )   http://taginfo-dev.opengeo
>>> data.hu/
>>> Source code / issues / problems /  latest info  =>
>>> https://github.com/taginfo/dockerized-taginfo
>>>
>>> ========
>>>
>>> I am believing that a customized local taginfo can help on HOT's
>>> disaster preparedness
>>>
>>> The global taginfo (https://taginfo.openstreetmap.org/ )   is a very
>>> good tool for find tagging trends,
>>> but shows all problems  in the OSM - and this is similar to the
>>> "diffusion of responsibility", too many problems.
>>> The local taginfo show only the local data and so much easy to detect
>>> tagging problems, bad imports,  and other data quality issues.
>>> But this is my theory, no proof yet.  And there are more dead taginfo
>>> projects than living, just see: https://wiki.openstreetmap.org
>>> /wiki/Taginfo/Sites
>>>
>>> So I have started an experiment and created a program to generate a
>>> local taginfo server configuration for every area ( ~ 4000 )
>>> And now I can run ~100 mini taginfo server on a cheap 15GB Cloud
>>> server.  ( but this is only a Proof of Concept, more work need )
>>>
>>> So If somebody wants to help on testing or just interested your area :
>>> the (temporary)  links here:
>>> http://taginfo-dev.opengeodata.hu/africa/  ( every African country ! )
>>> http://taginfo-dev.opengeodata.hu/central-america/  ( every
>>> Central-America country )
>>> ( Important:  temporary links for the next 2 weeks!  daily refresh only
>>> for Africa and Central-America  )
>>>
>>> If you need an extra country for testing - write me.
>>>
>>>
>>> For example:
>>>
>>> *Taginfo for Tanzania:*   http://af-tz.taginfo-dev.opengeodata.hu/
>>> * Amenity :  http://af-tz.taginfo-dev.opengeodata.hu/keys/amenity#values
>>>
>>> find undocumented key-values
>>> * amenity=mobile_money_agent
>>> * amenity=private_toilets   / amenity=private_toilet
>>> * highway=footpath
>>> * ...
>>>
>>> or find an undocumented keys or typos:
>>> * damage=             http://af-tz.taginfo-dev.open
>>> geodata.hu/keys/damage
>>> * blockage=            http://af-tz.taginfo-dev.openg
>>> eodata.hu/keys/blockage
>>> * mjumbe=             http://af-tz.taginfo-dev.open
>>> geodata.hu/keys/mjumbe
>>> * highways=           http://af-tz.taginfo-dev.open
>>> geodata.hu/keys/highways   typo 327 highway(s)
>>> * buildng=              http://af-tz.taginfo-dev.ope
>>> ngeodata.hu/keys/buildng       typo
>>> * top_width=           http://af-tz.taginfo-dev.open
>>> geodata.hu/keys/top_width
>>>
>>> checking tagging problems:
>>> http://af-tz.taginfo-dev.opengeodata.hu/reports/characters_in_keys#space
>>> http://af-tz.taginfo-dev.opengeodata.hu/reports/characters_i
>>> n_keys#letters
>>>
>>> Data cleaning:
>>> *  addr:city=*   http://af-tz.taginfo-dev.open
>>> geodata.hu/keys/addr%3Acity#values  ( check pages 3-  )
>>> *  building:material= http://af-tz.taginfo-dev.opengeodata.hu/k
>>> eys/building%3Amaterial#values
>>> *  boundary=    http://af-tz.taginfo-dev.ope
>>> ngeodata.hu/keys/boundary#values
>>>
>>>
>>> Other Examples:
>>>
>>> Nicaragua:   http://ca-ni.taginfo-dev.opengeodata.hu/
>>> Haiti:  http://ca-ni.taginfo-dev.opengeodata.hu/
>>> Cuba:  http://ca-cu.taginfo-dev.opengeodata.hu/
>>>
>>> Burkina Faso: http://af-bf.taginfo-dev.opengeodata.hu/
>>> Zambia: http://af-zm.taginfo-dev.opengeodata.hu/
>>> Niger: http://af-ne.taginfo-dev.opengeodata.hu/
>>> Nigeria: http://af-ng.taginfo-dev.opengeodata.hu/
>>> Kenya: http://af-ke.taginfo-dev.opengeodata.hu/
>>> Mali: http://af-ml.taginfo-dev.opengeodata.hu/
>>> Zimbabwe http://af-zw.taginfo-dev.opengeodata.hu/
>>> South Africa: http://af-za.taginfo-dev.opengeodata.hu/
>>>
>>> ....
>>>
>>> Some Extra Bonus  ( no daily refresh  , just a Proof of Concept  )
>>> ==============================
>>> California (US) : http://na-us-ca.taginfo-dev.opengeodata.hu/
>>> Mississippi (US) http://na-us-ms.taginfo-dev.opengeodata.hu/
>>> Berlin (DE):  http://eu-de-be.taginfo-dev.opengeodata.hu/
>>> British Columbia (CA):   http://na-ca-bc.taginfo-dev.opengeodata.hu/
>>> Istanbul (TR):  http://eu-tr-34.taginfo-dev.opengeodata.hu/
>>> Saint Petersburg (Russia) http://ru-ru-spe.taginfo-dev.opengeodata.hu/
>>>
>>> Mexico: http://na-mx.taginfo-dev.opengeodata.hu/
>>> Greenland: http://na-gl.taginfo-dev.opengeodata.hu/
>>> Columbia: http://sa-co.taginfo-dev.opengeodata.hu/
>>> Papua New Guinea: http://ao-pg.taginfo-dev.opengeodata.hu/
>>> Indonesia: http://as-id.taginfo-dev.opengeodata.hu/
>>> Sri Lanka:   http://as-lk.taginfo-dev.opengeodata.hu/
>>> Myanmar: http://as-mm.taginfo-dev.opengeodata.hu/
>>> Mongolia: http://as-mn.taginfo-dev.opengeodata.hu/
>>> Nepal:  http://as-np.taginfo-dev.opengeodata.hu/
>>>
>>> Antarctica:    http://aq-aq.taginfo-dev.opengeodata.hu/
>>>
>>>
>>> see all:  http://taginfo-dev.opengeodata.hu/
>>>
>>>
>>> Disclaimer:
>>> *This code is based on other open source projects:  see
>>> https://github.com/taginfo/dockerized-taginfo#credits  and using
>>> Geofabrik daily extracts.
>>>
>>>
>>> bonus for testing:
>>> =======
>>> *"cuisine"  * For describing the type of food served at an eating place.
>>> * Tanzania:   http://af-tz.taginfo-dev.open
>>> geodata.hu/keys/cuisine#values
>>> * California:  http://na-us-ca.taginfo-dev.opengeodata.hu/keys
>>> /cuisine#values
>>> * Mexico:    http://na-mx.taginfo-dev.opengeodata.hu/keys/cuisine#values
>>> * Cuba   :    http://ca-cu.taginfo-dev.opengeodata.hu/keys/cuisine#value
>>> s
>>>
>>> *"denomination"  *Defines the subspecies of a specific religion
>>> * Sri Lanka: http://as-lk.taginfo-dev.opengeodata.hu/keys/denomina
>>> tion#values
>>> * South Africa: http://af-za.taginfo-dev.opengeodata.hu/keys/denomin
>>> ation#values
>>> * Mongolia: http://as-mn.taginfo-dev.opengeodata.hu/keys/denom
>>> ination#values
>>> * Indonesia:  http://as-id.taginfo-dev.opengeodata.hu/keys/den
>>> omination#values
>>>
>>> //  Remainder: links valid for the next 2 weeks!  after shutdown or
>>> change.
>>>
>>> I love this community  :)
>>>   Imre
>>>   // ImreSamu
>>>
>>>
>>>
>>>
>>> _______________________________________________
>>> HOT mailing list
>>> HOT at openstreetmap.org
>>> https://lists.openstreetmap.org/listinfo/hot
>>>
>>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/hot/attachments/20180321/69fedb4d/attachment.html>


More information about the HOT mailing list