Please note that I do not attempt to classify images to all possible categories or by adding them to pages. I however ensure that they belong to at least one suitable category and include the description which has the keywords needed for a searcher to find the image. In all likelihood the image will also be in use on a relevant page on the English wikipedia.
It is particularly unfortunate that Commons / Mediawiki lacks the semantic plugin or other such alternatives - which can lead to a potential explosion of categories. For instance there is "Category:Bird family Y" and "category:Birds of Location X" and then there are passing users who create "category:Birds of Location X belong to family Y" - this is clearly non-scaleable. I think temporal, geographic, conceptual classes need to be kept apart and no combinations should be made (eg:prefer India, 1887 as two categories rather than India in 1887). Categories should be "atomic" - similar to Codd's rule.
I believe that spatial categories, especially as used for organisms, should not be done with named entities - we ought to have either bounding rects or coordinate polygons and all search should be based on a table of named entities and shapes. Definitely something for Wikidata to aim for.
- PZSL 1903 2
TUSC token: 01b673d92800a5e4ea8f6e6ca27b1da1
I am now proud owner of a WMFLabs TUSC account!