Platonides escribió:
Then the category tree is not too well organised. Having only categories on the first levels and then articles in then would be much more logical.
The issue with child categories is that the category organization is up to the wiki community. There may be loops. The articles may be both in the parent and child categories, it may have hundreds of descendants... ...and the operation would need to be efficient.
Platonides, as you said, the categories are a mess in the Wikipedia. The "Categorization" proyect just try to put, at least, one category holding each article, but the category tree is hand work of the wikipedians. There's no tree, there's a lot of orphans categories, duplicated ones, and a lot of loops and crosses.
That's because I assume a simple recursive depth search is the only solution for getting a random article in a category (or their childs)