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

Re: Versions, categorization, dmoz.org

Author:James Carlyle
Posted:9/9/1999; 7:02:05 AM
Topic:rss channels via email
Msg #:10814 (In response to 10803)
Prev/Next:10813 / 10815

Comments on SN 2.0b1: I wasn't suggesting a move to "2.0", just requesting clarification on where things stood. But if I want to store the DTD/schema URL for SN, what should I use? There used to be a dtd but it seems to have moved.

Comments on Dewey: It's not perfect, since it reflects to a certain degree the Christian Anglo-Saxon bias of Dewey himself. But they are working hard at that. It doesn't map the Internet and other 'latest' technologies well - of course changes to the structure are subject to a peer review (but this is a Good Thing, IMHO). But the important point to remember is that in many cases the Internet aspect is just the delivery mechanism (i.e. Internet Banking should be classified under Banking, not Internet), so this is less a limitation than it might appear.

The problems with using dmoz.org as a channel directory are:

Technical:

The current channel listings are not open - no-one outside Netscape has any way of knowing what the URLs of the listed channels are, only how to add them to My.Netscape (like http://my.netscape.com/addchannel.tmpl?service=net.331) At least My.Userland and xmlTree publish the channel URLs in both HTML and XML formats.

It (as a directory) is not tuned to the needs of channels, like when they were last updated, what format they use, who the contact email is, and so on. The directory is tuned to display just the link title, the URL, and a description. For example, we are creating a dumbed-down channel directory to meet the needs of channel users, hiding the complexity of XML (probably not even mentioning XML!), by adding only the listing details which might be useful.

It does not offer the opportunity to rebrand the channel directory content, except by using the 25 MB dump (in obsolete RDF) of the whole dmoz site. Compare that with http://www.takepart.com/container.cfm?DDC=005&L=3&Stylesheet=http://www.xmltree.com/myUserland.xsl (for those who missed my post yesterday).

There is no way of navigating the directory by XML.

Structural:

The directory structure itself is ad-hoc. The interpretation for how to categorise resources is left to each of the 10,000 editors - I can't find any published rationale for categorising in X vs. Y

There is no work done on building DMOZ --> Library of Congress or Yahoo classification conversion utilities.

The use of DMOZ requires English comprehension (c.w. Dewey, which can be used with any language)

There is no work done on automated or aided classification.

The system is not faceted.

You can't hire a catalog expert who understands the whole structure of DMOZ (not easily, at least).

There are responses to this message:

This page was archived on 6/13/2001; 4:52:31 PM.

© Copyright 1998-2001 UserLand Software, Inc.