Hey guys,

Thanks a lot for the replies. I might have been a bit vague what I meant by
automated:D So, let me clear that up. I plan to simply make a script to
recommend categories from an image. There would be users who do check if
the category is valid before adding it right now, as the script may not be
a 100% reliable (yet). I do understand about bots on Commons and I would
definitely not be running any of this without appropriate approval :) So,
currently we just recommend categories, for example you can check
https://commons.wikimedia.org/wiki/User:AbdealiJKTravis/logs/newimages

Also, Adding Category:Paris to an image is quite fine I believe, because
users do have the template {{OverPopCat}} to search for categories which
require diffusion. So, essentially if the script is able to add/suggest
Category:Paris, there would be users who are interested in Paris
specifically who will be able to find that image easily and add a deeper
category based on their knowledge. There's also {{check categories}} for
every file which helps in such cases.

I do agree I need a reverse geocoding service. I tried nominatim's reverse
querying, and was hoping there was an equivalent hosted by wikimedia.

Regards

On Wed, Jul 13, 2016 at 1:45 AM, T Fish <[email protected]> wrote:

> What you need is a decent reverse geocoding service.
>
>
> I agree with Tim's objection to automating this task. Also, the local
> community[1] needs to be consulted before running a task. I would
> suggest reaching out to them before starting any code.
>
> [1] https://commons.wikimedia.org/wiki/Commons:Bots/Requests
>
> On Tue, Jul 12, 2016 at 2:32 PM, Tim Alder <[email protected]> wrote:
> > Hello,
> > I don't believe it's an good idea to add automatically Paris as Category
> > for an image of the Eiffel Tower. Category should be a deep as possible.
> > Somebody has to clean-up this add will not be happy. Perhaps better to
> > have a maintainance category for this.
> >
> > Perhaps you can use existing images with GPS and Categories to add
> > categories for new images. (Sound like machine learning.)
> >
> > If you see a chance to link from Wikipedia/Wikidata to Commons, you
> > could use coordinates from Wikipedia (something I could provide).
> >
> > One other option for mayor cities could be to setup populate place from
> > naturalearthdata:
> >
> http://www.naturalearthdata.com/downloads/110m-cultural-vectors/110m-populated-places/
> >
> > Greetings Tim alias Kolossos
> >
> >
> >
> > Am 12.07.2016 um 05:25 schrieb Abdeali Kothari:
> >> Hi,
> >>
> >> I am a GSoC student making a script using pywikibot to find categories
> >> for Images on commons[0]. One thing I was trying was to use the GPS
> >> coordinateds given by the camera (in EXIF data) to find the location and
> >> add the place's category to it. For example, if I take an image of the
> >> Eiffel Tower it should auto add Category:Paris to it.
> >>
> >> For this, I was trying to use the Nominatim service from OSM.
> >> http://nominatim.openstreetmap.org is not meant for bulk finding the
> >> geolocations and limits the query rate at 1 query / minute.
> >>
> >> I was wondering whether the OSM api was also available
> >> on maps.wikimedia.org <http://maps.wikimedia.org/> ? And whether the
> >> Nominatim API is hosted and working ? I wasn't able to find any
> >> documentation about it.
> >> Also, is there any documentation on query limits and so on ?
> >>
> >> Regards,
> >> Abdeali JK
> >>
> >> References:
> >> [0] - https://phabricator.wikimedia.org/T129611
> >>
> >>
> >>
> >> _______________________________________________
> >> Maps-l mailing list
> >> [email protected]
> >> https://lists.wikimedia.org/mailman/listinfo/maps-l
> >>
> >
> >
> > _______________________________________________
> > Maps-l mailing list
> > [email protected]
> > https://lists.wikimedia.org/mailman/listinfo/maps-l
>
> _______________________________________________
> Maps-l mailing list
> [email protected]
> https://lists.wikimedia.org/mailman/listinfo/maps-l
>
_______________________________________________
Maps-l mailing list
[email protected]
https://lists.wikimedia.org/mailman/listinfo/maps-l

Reply via email to