Archive of UserLand's first discussion group, started October 5, 1998.

Why not have a patchwork quilt?

Author:James Carlyle
Posted:9/10/1999; 1:06:34 AM
Topic:rss channels via email
Msg #:10855 (In response to 10830)
Prev/Next:10854 / 10856

Jon

I actually like the idea of a patchwork quilt. You suggested:


...
...
OpenSource
Programming/Python
WebApplicationServers/Zope
Databases/OODB
...

I do not believe that the Dewey System should be the only way for finding channels, and xmlTree has two other distinct navigation mechanisms. One is a standard free-text search on the channel description, title, organisation name etc. The other uses keywords, like atomic versions of the categories you have suggested above, except that the keywords currently describe each channel and not each item.

The interface lists the 100 most popular keywords. The user chooses one. The channels that have that keyword are displayed, and the list of the 100 most popular keywords that those channels have in common is now displayed. The user picks more keywords, and the number of channels gradually narrows. The user can realise that an earlier keyword choice is no longer relevant, and unpick it, opening up the selection of channels.

Since a picture is worth a thousand words, you can see what I mean at http://www.xmltree.com/metadata/search.cfm

So the user moves around a 'cloud' (i.e. non-hierarchical structure) of keywords, and can arrive at a channel by an infinite number of different routes. Compare this to the directory structure, where one false turn two moves ago means that you will never find what you're looking for.

For example, I thought 'I wonder how easy it is to find Scripting News'. I selected 'news' (it made the top 100). I then selected 'scripting' (it made the top 100 after 'news' had been selected'). And I got five matches, including SN and Python Projects. As an aside, if I relaxed 'news' next, I would get 7 matches, including 'List of site changes at www.scripting.com'

There is no central classification body. The overhead for building the classification necessary for navigation is minimal. You do not need to be trained to classify, and hopefully using it is easy, to.

There are responses to this message:

This page was archived on 6/13/2001; 4:52:32 PM.

© Copyright 1998-2001 UserLand Software, Inc.