[Peerpress-main] Re: [IMC-Tech] categorization & editorial policy

[Joakim Ziegler]

On Tue, Jan 23, 2001 at 01:40:24AM -0500, Josh Marcus wrote:
> This conversation about how to organize and share independent
> news and opinion is very exciting to me (whoo!) -- I hope this conversation
> will continue until we come up with a fair and functional plan.

Since this discussion is very closely related to the taxonomical hierarchy
work I'm currently doing for Peer Press, I thought I'd outline in what
direction we're going with this sort of work.

A given Peer Press story has several categories it fits into. This is a must.
There's no way you're going to design a taxonomy that will allow you to fit
every article neatly into one category while not belonging in any other.

Additionally, we're currently looking at two taxonomy trees. One is the
topical, which you're discussing here, and the other is the geographical,
which is in the form of Planet / Continent / Subcontinent / Country / State
or Province and so on. Planet might seem silly, but we actually want to carry
astronomy news.

As for the topical taxonomy, I've been analyzing the topic hierarchies used
by large content providers, and I'm rather close to synthesizing at least two
levels of general topical hierarchy from that. The category list that was
posted here is too specific to fit into any of those two levels, although it
would most likely belong comfortably on the third level. Top level categories
are along the lines of Headline News, Politics, Technology, and so on. Second
level categories would include Politics/Activism, for instance, where it
seems to me a lot of your categories would fit nicely.

The Peer Press topic hierarchy will most likely have no more than four
levels. If it has more, it will become extremely wide, and will be more along
the lines of a keyword search. 

However, we're currently thinking about creating a third taxonomy (or
flagging some nodes in the topical hierarchy), for holding dynamically
created branches. Realise that it's very important to not make the hierarchy
totally dynamic. That almost completely negates its usefulness, especially in
the realm of automated filtering of content. If you don't know that a
category is there, you can't set up your scripts to look for it.

On the other hand, dynamically generated branches are useful to group
articles about the same thing, to use for "Show me more articles like this
one" and other human intervention-driven browsing. A very relevant use for
dynamically added categories would be a "Current events" branch of the
taxonomy.

So these are the general thoughts I've made so far about information
taxonomies. Beware, this is rather tricky territory, and difficult to do
right. We'd be wise to pool our resources and create a single standard
taxonomy for news content, since there's no such standard at the moment, at
least not a well-documented free one.

-- 
Joakim Ziegler - Ximian web monkey - joakim at ximian.com - Radagast at IRC
  FIX sysop - free software coder - FIDEL & Conglomerate developer
         http://www.avmaria.com/ - http://www.ximian.com/