[Talk-ht] incorrect health facility imports

Kate Chapman kate at maploser.com
Mar 10 Avr 19:09:33 BST 2012


Hi All,

Why not remove the ones that are not there?  They can always be added
back if someone does locate them.

This original dataset was imperfect and expected to improve overtime,
but didn't really completely finish before the "final release."

Part of updating OpenStreetMap data and surveying should be removing
data that doesn't exist as well.  I would think that this is also the
case with some of the IDP camps as well.

Best,

-Kate

Salut à tous,

Pourquoi ne pas supprimer celles qui ne sont pas là? Ils peuvent
toujours être rajoutés si quelqu'un ne les localiser.

Cette base de données originale a été imparfaite et devrait permettre
d'améliorer les heures supplémentaires, mais il n'a pas vraiment
complètement terminé avant que la «version finale».

Une partie de la mise à jour des données OpenStreetMap et d'arpentage
devrait être la suppression des données qui n'existent pas ainsi. Je
pense que c'est aussi le cas avec certains des camps de personnes
déplacées ainsi.

Best,

-Kate

On Tue, Apr 10, 2012 at 1:00 PM, Brian Wolford
<worldwidewolford at gmail.com> wrote:
> I see the point of the features still being useful when they can be
> relocated. But as we are moving towards making final end products with OSM
> data these "pollutants" are becoming more of nuisance. For instance the maps
> that are being made by AFH right now are having these random facilities come
> up on the map (these facilities were not found on the survey of the area
> either). And while these can be removed on the map making/GIS side (and I
> don't believe in punishing OSM with GIS woes) the maps are made to be easily
> updated from OSM data. Which makes that more difficult. And I feel that
> using mapnik rendering tags to designate symbols later should not be
> something that needs workaround.
>
> Another, potentially dangerous, thing I see from leaving rendered health
> facilities in known incorrect spots (this might be a bit of a stretch
> currently) is people using navigation devices (or even paper maps). If there
> were some one out in a rural area who was injured and used their Garmin
> loaded with OSM data to find the nearest health facility, they might be
> directed to one of the known incorrect locations and have put them self even
> farther from medical attention.
>
> I would rather, and have been, remove the 'amenity=' tag from these while
> they are known incorrect and remove them from the rendering. They can always
> be found later using a location fixme tag like Severin suggested.
> I personally don't think that is even needed tho. If one would like to find
> these features you can filter for the paho tags. Or if you'd really like to
> work on them click the JOSM link on this page:
> http://taginfo.openstreetmap.org/keys/paho%3Aid
> They will not be unfindable without the amenity=hospital tag.
>
> I only suggest removing tags rendering tags when they are known incorrect
> and basic research does not reveal a better location.
>
> Retorts?
> -brian
>
>
> On Tue, Apr 10, 2012 at 9:09 AM, Jaakko Helleranta.com
> <jaakko at helleranta.com> wrote:
>>
>> Desole seulment en francoise. Svp utilise http://translate.google.com si
>> tu comprends pas l'anglais.
>> --
>> (This is a few days old by now as the reply didn't get sent before the
>> other replies but these r my thoughts anyways.)
>>
>> I see this being somewhat closely related to a wider problem about adding
>> crisis response data / less-than-desirable-quality data into OSM during
>> after acute crisis.
>>
>> On one hand it's purely against OSM import rules.
>> One the other hand, Haiti was the first time such broad use of OSM
>> happened in a such a big crisis. .. And in an area where the original map
>> was practically non-existent.
>>
>> The problem has different perspectives to it of which I can think of two
>> top issues:
>>
>> * it was more or less an emergency import (which doesn't mean that we
>> can't clean it)
>>  - due to this there's a _ton_ of emergency health facilities (field
>> hospitals, etc). These I tend to delete unless there's a reason to believe
>> that there may be something there, of course.
>>
>> * the health facility imports done were the "best" data that was
>> available. You can find more or less the same data in the Government's Carte
>> Sanitaire (Haiti Health Map, http://j.mp/cartesanitaire if I remember the
>> shortlink right) (which is no excuse to dump crappy data to OSM...)
>> This includes hospitals in the middle of the ocean, which I've usually
>> dragged closer to the shore :) ...
>>
>> Now what to do with these you ask?
>> Good question.
>>
>> I personally haven't removed the amenity=hospital tags even though they
>> _are_ polluting the map as you say.
>> When trying to improve things by my self I've tried to move nodes closer
>> to what I'm guessing may be the place where the health facility is -- if I
>> have any clue.
>> The thing is that there may very well be _some_ heath facility in the area
>> as I have found out by asking about the existence / locations of a number of
>> facilities from health sector people I know.
>> I've adjusted a number of health facility locations with locals / people
>> with local knowledge and this has usually been very ad hoc "let's look at
>> the map since you know the area and you can tell me if these facilities
>> exist somewhere in the area so we can adjust the locations and information".
>> If the amenity=hospital tag is removed this kind of fixing can not happen.
>> .. Also, deleting the "main tag" would simply hide the problem and
>> significantly(?) increase the risk of someone entering a new node of the
>> facility and nobody spotting the duplicate afterwards.
>> So, rather than deleting the main tag I try to tag them with "FIXME=survey
>> location/existence" or something alike.
>>
>> I guess I adhere here to the though of: "Better roughly right than
>> precisely wrong -- or not there at all" even though the health facilities
>> (including some of the by-now-non-existing cholera treatment facilities,
>> btw!) are polluting the map and are annoying...
>>
>> As you can see, I have no clear solution/suggestion to the problem.
>>
>> Thoughts, ideas, suggestions, anyone??
>>
>> Cheers,
>> -Jaakko
>>
>> P.S. This issue relates also to other quake response data added en-masse
>> to OSM soon after the quake: I've recently nuked a ton of collapsed building
>> nodes (a surprisingly large share of which were not collapsed to begin with,
>> btw -- just normal Haitian unfinished constructions looking like "collapsed"
>> for unexperienced remote mappers..). There's similar issue with "mudslides"
>> (being in reality just typical soil erosion), etc.
>> P.P.S. HOT list has an partly related active topic on the issue of natural
>> hazards in OSM.
>>
>> Sent from my BlackBerry® device from Digicel
>> --
>> Mobile: +509-37-26 91 54, Skype/GoogleTalk: jhelleranta
>>
>> -----Original Message-----
>> From: Brian Wolford <worldwidewolford at gmail.com>
>> Date: Thu, 5 Apr 2012 10:03:40
>> To: Talk-HT at OSM<talk-ht at openstreetmap.org>
>> Subject: [Talk-ht] incorrect health facility imports
>>
>> _______________________________________________
>> Talk-ht mailing list
>> Talk-ht at openstreetmap.org
>> http://lists.openstreetmap.org/listinfo/talk-ht
>>
>
>
> _______________________________________________
> Talk-ht mailing list
> Talk-ht at openstreetmap.org
> http://lists.openstreetmap.org/listinfo/talk-ht
>



Plus d'informations sur la liste de diffusion Talk-ht