[OSM-talk] [Talk-us] Redacting 75, 000 street names contributed by user chdr

James james2432 at gmail.com
Sun Aug 27 19:38:34 UTC 2017


The odbl=clean could be a way to exclude road names from the removal,
though sources from the info should be given, whether Canvec, Mapillary,
OpenStreetCam or local survey

On Aug 27, 2017 3:29 PM, "john whelan" <jwhelan0112 at gmail.com> wrote:

> There are a couple of issues here.  The first are our users, we don't
> normally think about them but deleting the names at the wrong point in
> OSMAND's cycle could mean missing street names for a period of time.
>
> Second is the problem of some data might be incorrect as a result of the
> source deliberately using invalid names.  The very clean way is to delete
> then retag.
>
> Verifying with a maproulette challenge would work well if we could trust
> all the mappers not to just tick the box either deliberately or by mistake.
>
> Since we have the location of the streets and we have other sources with a
> valid name which would probably vary by country could someone join the two
> together and verify the name in an automated way?  Leaving a much smaller
> list of street names to be verified manually?  I suspect Jamie could wave a
> magic wand for Quebec.
>
> I'm not saying we should do one thing or another here.  I'm attempting to
> analyse the problem and find a solution that impacts as few people as
> possible but gives us clean accurate data at the end of the process.
>
> If we go the verifying route could we take a page out of HOT's process and
> have someone verify them someone validate?
>
> Cheerio John
>
> On 27 Aug 2017 3:04 pm, "Nicolás Alvarez" <nicolas.alvarez at gmail.com>
> wrote:
>
>> I don't understand what people mean with 'verifying' objects. We're
>> not trying to find factually-incorrect data. The data is legally
>> tainted. It's questionable whether looking at the current names
>> imported from GMaps, comparing to another source, seeing they match
>> and marking them as "verified" will legally change anything. And it's
>> impossible to know if people are really verifying anything or just
>> blindly marking them as verified.
>>
>> I think the only clean way to solve this is to redact and then re-map
>> from legal sources.
>>
>> --
>> Nicolás
>>
>> 2017-08-27 14:39 GMT-03:00 Frederik Ramm <frederik at remote.org>:
>> > Steve:
>> >
>> > thank you for your work. I'll save your list. It appears that others
>> > might be eager to do the same, maybe we can find a good workflow for
>> > that. I wasn't expecting the community to start working on this
>> > pre-redaction but if people prefer that to fixing issues later, it is of
>> > course an option. I certainly prefer out-of-band "marking" of verified
>> > objects to adding a new tag to each!
>> >
>> > Tod:
>> >
>> > On 08/27/2017 07:31 PM, Tod Fitch wrote:
>> >> When you reviewed Orange County, how did you do it so quickly? The
>> only way I know to go through this is looking at each one, one at a time.
>> >
>> > I could of course make a page with links to the ways, even per county if
>> > that helps, or we could upload the list to some suitable tool. Ian
>> > mentioned MapRoulette but I'm not sure if that would make things easier.
>> > I'm certainly happy to try. Maybe Martijn would like to chip in about
>> > MapRoulette?
>> >
>> > Bye
>> > Frederik
>> >
>> > --
>> > Frederik Ramm  ##  eMail frederik at remote.org  ##  N49°00'09"
>> E008°23'33"
>> >
>> > _______________________________________________
>> > talk mailing list
>> > talk at openstreetmap.org
>> > https://lists.openstreetmap.org/listinfo/talk
>>
>> _______________________________________________
>> talk mailing list
>> talk at openstreetmap.org
>> https://lists.openstreetmap.org/listinfo/talk
>>
>
> _______________________________________________
> talk mailing list
> talk at openstreetmap.org
> https://lists.openstreetmap.org/listinfo/talk
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk/attachments/20170827/b24170b1/attachment-0001.html>


More information about the talk mailing list