[Tagging] Automated edit of image tags suggestion

[pangoSE]

thanks for the constructive suggestions :)

Den Thu, 25 Jun 2020 22:14:21 +0200
skrev Re: [Tagging] Automated edit of image tags suggestion:

> sent from a phone
> 
> > On 25. Jun 2020, at 19:59, pangoSE <pangose at riseup.net> wrote:
> > 
> > image=File:* -> commons_file=File:* image=Category:* ->
> > commons_category=Category:*
> > image=https://commons.wikimedia.org/wiki/File:* ->
> > commons_file=File:*
> > image=https://commons.wikimedia.org/wiki/Category:* ->
> > commons_category=Category:*  
> 
> 
> splitting commons into files and categories (different keys) seems to
> be an improvement, although neither of these keys are existing at the
> moment. Following what we have so far, “wikimedia_commons_file” and
> “wikimedia_commons_ category“ would fit better, although a bit
> unwieldy.

Fixed in algorithm 2 in the wiki.
https://wiki.openstreetmap.org/wiki/Automated_edits/pangoSE#Key%3Aimage

> 
> From a datauser perspective, does it really improve the situation?

Yes. Commons image-urls can be specified so that a certain size is
returned. If the commons-url is in the image-tag I have to first match
it with a regex, which is an extra step. Some image tags have multiple
urls also which also has to be dealt with, etc. Its a mess from a
dataconsumer viewpoint. 
I experienced it myself making this:
https://github.com/pangoSE/sheltermap

> Right now you have to check for 2 “main” keys: image and
> wikimedia_commons (leaving wikidata out for the moment), and then you
> can see what you find in the value (url, file: category: etc.) after
> your proposed edit you would have to check for more keys but could
> hope that the values would be better standardized. And you’d have to
> run a bot frequently to keep things “clean”.

Is there anything preventing us from running bots (with simple
algorithms) on the database? Wikimedia projects do that all the time. I
rarely see this in OSM (besides the http/https bot)

> 
> Btw, there are also a few images tagged with a “flickr” key (~1200)
> While it could eventually make sense to make an exception for
> wikimedia commons, I do not believe we should create a new key for
> every image hosting service.

For me there are 2 categories: sites hosting free images like
Flickr, Mapillary and Commons and all the rest. All the rest can be in
the image tags. The 3 before mentioned should be kept in their tag if
not for other reasons for statistical purposes.

I'm not advocating creating any more tags for other services here.

Cheers pangoSE