Redundant Categories

I have noticed that many general categories are very much redundant. For example, how many News & Media categories and subcategories do you need?

News & Media does not belong as subcategories hundreds of times.

Just my 2 cents.

Max

Well, News&Media is what we call a "template category". There are Newssites on pretty much every topic we have a category for. If there are enough of those sites for one single topic, we create a subcategory "Topic/News&Media". What should be wrong with that approach?

Remember that each category is qualified by the whole category path, so there should not be much redundance there.

News & Media does not belong as subcategories hundreds of times.

Click to expand...

You're absolutely right. Hundreds of times would be much too few. That's why we have thousands of them.

A moment's thought -- or a week spent studying ANY large taxonomic scheme -- will convince you that the phenomenon you describe is inevitable.

We have categories News, Tourism, Maps, and Business.

We also have categories Florida, Georgia (2!), and Bulgaria.

Where shall Bulgarian websites go? Should it be Business/Bulgaria, News/Bulgaria, Maps/Bulgaria (so Bulgaria is the "repeated but decidedly NOT redundant topic) or should it be Bulgaria/News -- so that there is also a Florida/News, Asia/Georgia/News, US/Georgia/News, ... and News is the repeated topic.

Whichever way you do it, your structure is easier to use if the subtopic names are consistently applied, and so taxonomists invariably produce "standard subtopic templates" to help them avoid unnecessary inconsistencies.

These orthogonal taxonomic hierarchies appear all over the ODP. Regionalization forms one obvious orthogonal taxonomy; media categorization (Magazines, Personal Pages, Directories, Sound Files) often forms one; perspective (religious sect, political slant, idiosyncratic scientific ideas) forms yet another. Language forms a slightly less obvious one. Some of these are ubiquitous throughout the ODP; in addition, very specific categories often have their own private orthogonal subcategorizations.

The ODP, the Dewey Decimal System, the Library of Congress cataloging system, all include MANY examples of "subcategorization templates". Yahoo has regional topics (perhaps after language the most pervasive of all orthogonal categories) built in as as a single orthogonal categorization -- so that there is really only one listing for, say, the Chicago Symphony Orchestra -- which appears whenever you drill down into any region containing Chicago deeply enough to see the musical ensembles. So in effect their top-level categories comprise their community templates, -- or their geographic structure defines a regional template: whichever way you're not looking at the directory. But they still have other duplicated category names for other kinds of orthogonal categorizations.

This isn't rocket science, but it is "Library Science 101."

Ok, I take it back. I hope I did not offend.

I'm always happy for an excuse to explain how a knowledge tool works. And a directory is a very powerful tool -- if you know how to use it, which is not always as obvious as one would think.

What does 'orthogonal' mean in this context? Mirriam-Webster wasn't much help here.

Nareau

m-w.com][b]5 :[/b said:

statistically independent

Click to expand...

searchStorange.com]In geometry said:

meaning angled). The term has been extended to general use, meaning the characteristic of being independent (relative to something else). It also can mean: non-redundant, non-overlapping, or irrelevant.

Click to expand...

I think "independent" and "non-overlapping" are the relevant terms.

If I knew a shorter word that meant the same thing, I'd have used it. (really!)

Two partitions of a set are orthogonal if their cross product is at least potentially fully populated.

Another way of saying it is that if you've determined which of one set of subcategories a site goes in, you really don't automatically know which of another ("orthogonal") of subcategories it goes in.

hutcheson said:

If I knew a shorter word that meant the same thing, I'd have used it. (really!)

Click to expand...

Ya, sure.

What does 'orthogonal' mean in this context?

Click to expand...

I like to think of it as a table or a matrix, with rows and columns (the rows and columns are orthogonal, or at right angles, to each other.) For example all business with a physical location have two basic orthogonal properties - location and topic (what business they are in.) You could create a table with locations across the top and business topics down the side and then each and every business would fit into one specific box which represented the proper location and topic.

In reality the problem is multi-dimensional (or in plainer speak - there are more than just two concepts to classify a website by.) The specifics about how to classify a website (e.g. London, Clothing Store, Online Shopping) is sometimes called "meta data" (meaning data about the data.) If the ODP were a real database instead of flat files we could do a lot more with using the meta data to create logical category structures rather than listing sites multiple times and linking categories together.

Hopefully I didn't confuse the issue further.

lissa said:

Hopefully I didn't confuse the issue further.

Click to expand...

Not at all. Very good descriptive definition.

hutcheson said:

I'm always happy for an excuse to explain how a knowledge tool works. And a directory is a very powerful tool -- if you know how to use it, which is not always as obvious as one would think.

Click to expand...

I know I agree.

I foound several very awsome link partner sites using the "search" feature :d

Max