Background Pony #7C7A
[@UnderwoodART](/forums/tagging/topics/general-tag-discussion?post_id=5235122#post_5235122), [@ZONESS](/forums/tagging/topics/general-tag-discussion?post_id=5235266#post_5235266), [@Background Pony \#52BA](/forums/tagging/topics/general-tag-discussion?post_id=5235299#post_5235299)
Tagging all facts or only unexpected facts? Should "frog" mean the animal or the part of a hoof? These are inherent problems of simple, text-based tags. To avoid them, we could use something with more structure. For example Wikidata has separate items for [frog the animal](https://www.wikidata.org/wiki/Q3116510) and [frog the hoof part](https://www.wikidata.org/wiki/Q5505173). Similarly, it knows that [Glock 17](https://www.wikidata.org/wiki/Q172172) is a [Glock](https://www.wikidata.org/wiki/Q739882) and a [pistol](https://www.wikidata.org/wiki/Q1907525), so you don't need to tag "Glock 17", "Glock" and "pistol" on the image, you just tag Q172172 (Glock 17). This also allows for tags to be translated to different languages while still being the same tag.
By the way, Wikidata already has some pony stuff modelled; see e.g. [Fluttershy](https://www.wikidata.org/wiki/Q5409355).
This would allow us to store default colors, features, etc. for characters in their respective items. There would be no need to tag them on each image, and yet it would still be possible to search for e.g. pictures of characters with blue manes. You could even search for characters born in Cloudsdale if such information was available for them.
Tagging all facts or only unexpected facts? Should "frog" mean the animal or the part of a hoof? These are inherent problems of simple, text-based tags. To avoid them, we could use something with more structure. For example Wikidata has separate items for [frog the animal](https://www.wikidata.org/wiki/Q3116510) and [frog the hoof part](https://www.wikidata.org/wiki/Q5505173). Similarly, it knows that [Glock 17](https://www.wikidata.org/wiki/Q172172) is a [Glock](https://www.wikidata.org/wiki/Q739882) and a [pistol](https://www.wikidata.org/wiki/Q1907525), so you don't need to tag "Glock 17", "Glock" and "pistol" on the image, you just tag Q172172 (Glock 17). This also allows for tags to be translated to different languages while still being the same tag.
By the way, Wikidata already has some pony stuff modelled; see e.g. [Fluttershy](https://www.wikidata.org/wiki/Q5409355).
This would allow us to store default colors, features, etc. for characters in their respective items. There would be no need to tag them on each image, and yet it would still be possible to search for e.g. pictures of characters with blue manes. You could even search for characters born in Cloudsdale if such information was available for them.