[OSM-talk] Proposed automatic replacements of multiple surface values - the third edition (review welcomed!)

[facebook_140f8d4e-9d8f-4d51-a5a7-320f53afc591 at vollbio.de]

Quoting collapses ... so I ll answer directly.

You asked: Would it make sense to document any of ... And yeah, I would
document these and put them into "material=" for the record.

If something in a foreign language is "paved" then they missed the point
of OSM and its standards. Best one can do, be sceptical, keep it on
record but downgrade it and hope some one encountering the wrong
designation will correct it. Just my opinion on a workable solution.

Yes, you are right "sett" would be more appropriate. I have barely seen
actual cobblestones in Germany.

On the reruns I don't really have an opinion. You will probably already
correct many instances. No idea whether it makes sense to actually
rerun, or just wait for the next editor to do this job.

Cheers, Klaus


On 25/03/2023 09:28, Mateusz Konieczny via talk wrote:
>
>
>
> Mar 24, 2023, 22:33 by
> facebook_140f8d4e-9d8f-4d51-a5a7-320f53afc591 at vollbio.de:
>
>     Hi
>
>     Thanks for the comprehensive explanation. Here some answers:
>
>      1. This first part makes sense to me.
>
> Thanks!
>
>      2. Regarding your questions to document previous values, I would
>         propose to use "material=..." as mentioned under:
>         https://wiki.openstreetmap.org/wiki/Key:surface#See_also
>
> which "previous values" you mean?
>
>      3. "tierra" and "terra" ... the answer is obvious: "ground".
>      4. Also, for other languages, which are not clear, I would use
>         the "material=..." key and just put "surface=unpaved", to be
>         on the safe side.
>
> If unclear, I will just not touch it (especially as some unclear may
> describe paved surface)
>
>      5. Btw. I saw some German word: holz=>wood, schotter=>gravel,
>         verdichtet=>paved (it means "compacted"), Gras/Laub=>grass,
>         pflasterstein=>cobblestones
>
> Could surface=verdichtet be actually surface=compacted?
>
> Is surface=pflasterstein a surface=cobblestone (underdefined one) or
> can it be assumed to be in every single case surface=sett or
> surface=unhewn_cobblestone ?
>
> See https://wiki.openstreetmap.org/wiki/Tag:surface%3Dcobblestone
>
> I added
>         # translating German
>         'holz': 'wood',
>         'schotter': 'gravel',
>         'Gras_Laub': 'grass',
> to candidate list for the next one (I will also review is any of them
> typically wrong
> - using aerial images - and is connected to any obvious issues not already
> automatically detectable)
>
>      6. Regarding regular reruns of this, I would put some quality
>         gates into that, like:
>           * Were previous automatic corrections reverted and why?
>           * Does a certain type reoccur often and why?
>           * Do new categories appear and should they become official?
>
> I inspect some sample of bot edits and before I run it I get info how
> many will be changed
> - so in case of large volume of cases (re)appearing I would
> investigate what is going on.
>
> I am also monitoring PMs and changeset comments on bot account - and I
> think
> that someone thinking that it made a bad edit would write.
>
> Does it seem enough for you? (I can also run this edit once and once
> enough new
> or repeated cases gather then appear here again with similar listing)
>
>
>
> _______________________________________________
> 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/20230325/a17651b7/attachment-0001.htm>