[WikiEN-l] Re: A first encounter with Categories

[Bryan Derksen]

At 03:22 AM 6/8/2004 +0200, Anthere wrote:
>[[list of museums in Australia with a chinese flag hanging above the 
>entrance door]] is of no use.
>
>Search "[[museum]]" where "[[country]] is 'australia'" and "[[topic]] is 
>'China'" is useful.
>
>To be useful, meta-tags must be large rather than very specific. Query 
>gives the specificity.

But thanks to subcategorization, they can be _both_. [[Museums in ohio]] 
can be a subcategory of [[Museums]] along with all the other [[Museums in 
<location>]] subcategories, and then all you need to do is throw in a 
mechanism for referring to all the articles in a category tree and you can 
treat [[Museums]] as if it contained all museum articles everywhere. If 
you're already proposing a query language for manipulating categories this 
seems a fairly trivial extension to it to me.

I've been wavering a bit, but I think I'm coming down pretty solidly in the 
fine-grained categorization camp. The other day I was working on moving the 
articles in the [[North American Rivers]] category to the [[North American 
rivers]] category (note capitalization), and started putting the articles 
into [[Idaho rivers]], [[Alberta rivers]], etc subcategories instead 
because the list was so big and generic otherwise. I then found it 
trivially easy to stick those subcategories under the relevant geographic 
categories as well ([[Idaho]], [[Alberta]]), and it occurred to me that it 
would be fairly easy once I was done to set up [[Mississippi watershed]], 
[[Colorado watershed]], etc. since in many cases whole entire states drain 
into those and I could just drop their subcategories in place to include 
all of their rivers. It would be a whole lot more work if I had to go 
around to each of those articles and add a new category to every river in 
North America.

IMO, fine-grained categories provide a convenient "handle" by which groups 
of articles can be organized in various useful ways, without requiring 
editors to learn how to deal with query languages doing complicated or 
mysterious stuff "behind the scenes" or adding lots of high-level category 
tags to individual articles.