Talk:Categorization with field-value pairs

Add topic
Active discussions

moved from article by Evan


The Nature of "Categories"

"Category" can be interpreted as a very broad concept. If considered as a set, its meaning is so broad that it is not a useful to the Wikipedia. Should a different, more specific term be used, or should the meaning be clarified so that it is merely jargon? Can the latter happen in a consensual process which is likely to involve directly or indirectly people who don't want it to succeed?

This is too much angels-on-a-pinhead for me. I think it will be a matter of practice how this feature is used. If you feel the need, change the fieldname "category" to some random character string. What's described is what most people seem to want in the categorization requirements.

Secondly are there ways of grouping articles that rely on the content of the articles in a way that is more part of the prose (like wiki and inter-wiki links)? If so, is this mechanism still useful?

Interlanguage links aren't part of the prose. (Interwiki links are, though). Why is this desirable? If you think that's a requirement, put it on categorization requirements, not here. --Evan 17:47, 18 Dec 2003 (UTC)

My only concern with this article is that it is very implementation-specific. Is there an article that discusses the theory underlying the implementation? Such an article would discuss ways of developing categories in general, perhaps with a theory and/or an algorithm for producing a categorized set of things (in this case, articles, but it does not have to be), not tag structures and other specifics. Brent Gulanowski 23:47, 16 Dec 2003 (UTC)

There's an article called Categorization which might be a good starting point. I think any theory about how to categorize things doesn't really pertain to new features for MediaWiki, though. --Evan 20:28, 17 Dec 2003 (UTC)

Categories and Subjects, and Articles and their Parts

What is the difference between stating the article w:History of Canada as part of w:Canada and putting it into the category of w:Canada?Will wikipedia users understand it?.

I think the idea of the part-of and is-in-category relationships is that parts are ordered. There's a possibility of different presentation, with Next, Prev, Top links or something similar. There's a proposal for doing this for Series of articles. It especially benefits Wikibooks, but there are parts of Wikipedia that could benefit, too.
Consider the w:Timeline of Quebec history. It's currently broken into parts, each of which covers a specific period of Quebec history. That would be a good candidate for a part-whole relationship.
Probably the biggest difference is that part-whole relationships relate articles, while categories relate subjects. Yes, it's a little abstruse. --Evan 18:57, 12 Dec 2003 (UTC)
An I thought articles deal with subjects (apart from disambiguation pages) ;-) Your right that next/prev-relations can be related to hierarical ones but that is not a must.
Yes, but the article is not the subject. Articles have properties that the subject does not. en:Nicole Kidman is 3500 bytes long, but Nicole Kidman is not 3500 bytes long. en:Nicole Kidman is available for free under the GFDL; Nicole Kidman is not available for free under the GFDL. --Evan 00:26, 16 Dec 2003 (UTC)

Problems with the Proposed Mechanism

When in article w:William Carlos Williams there is a link

 [[category=American poets]]

Why do you have to add in w:American poets something like


And what happens if it is missing?

The main reason is a technical one -- it's easiest to grab all the field-value pairs for a single article, and then run through those and use them in different way. If the "type" field for an article matches "category", then we go find all the articles that define it as a category, and lay them out nicely.
If instead, for each article, we checked to find any other article that has a "category" field that has this article's title as the value, that's an extra database call for every article -- most of which will come up empty. So, it's wasteful.
That said, I agree that having to add the "type" field is a problem. If it's missing, an article that other articles refer to as their "category" wouldn't actually list those articles out.
I'll add this as a disadvantage. --Evan 18:57, 12 Dec 2003 (UTC)
Why not make it impossible for the type field to be missing? In other words, such meta-data is not editable by contributors. (More detail now on categorization requirements. Brent Gulanowski 05:08, 19 Dec 2003 (UTC)

Unique identifiers distinct from names

I added this idea to the categorization requirements page, but thought I'd mention it here. What about using an ID for categories distinct from the descriptive name? Names can, sometimes should, change all at one go, whereas membership in a category would change one by one. In terms of tags: [[category=00323502]], and there would be a page with [[type=category key=00323502 value="Beat Poets"]] -- that is, the article which discusses the category of beat poets, and which might be renamed safely in future. Then you have some link which generates the list of articles with the 00323502 category (in the event you don't want to have summary article and membership list on the same page). Brent Gulanowski 05:08, 19 Dec 2003 (UTC)

Return to "Categorization with field-value pairs" page.