On Mon, 08 Nov 2004 14:48:48 -0800, Tim Bray <[EMAIL PROTECTED]> wrote:

> My findings:
> - more than half have no category information at all
> - of those that do, almost all are using dc:subject
> - I saw exactly *one* instance of RSS's <category> element, in feeds
> produced by Roller

Thanks, I was wondering.

> So... options are:
> 
> 1. put in category somewhat as I proposed it, on the grounds that it's
> been specified before and it's fairly clean
> 2. don't put in category, on the grounds that it's not been deployed
> 3. don't put in category but recommend the use of dc:subject

Of these options I'd lean closest to 1. but advocate a little further
exploration of what's needed. I like the idea, and <category> seems a
reasonable name.

I'd express the requirements as follows : whatever syntax is used it
should include at least a URI for the scheme, a place for an
identifier of the category within that scheme and a string for a
human-readable label. It should be optional. It should be possible to
use it at either feed or entry level.

As you say, dc:subject is in use (which discounts 2.), but I also
think this is something people would use more if it was reasonably
well worked-out and standardised.

I don't actually think dc:subject as it is currently used is tightly
constrained enough for what's needed - it trades broad deployment for
precision. Atom's element isn't likely to be used that much outside of
Atom.

I don't think RSS 2.0's approach is actually far off the mark, it is
pretty simple and supplying the URI of the schema in use is a major
plus for using alongside other (meta)data systems. But the slash
syntax seems an awfully clunky approach in a syntax that has pretty
natural support for ordered lists and hierarchies... [someone got
shares in MicroParsers Inc..?]

Putting the old RDF hat on, I reckon the SKOS work [1] will mean it
shouldn't be too difficult to map to other systems/schemes as long as
whatever Atom does is reasonably unambiguous.

fyi, there's some other plain-XML prior art that is been used
pointillistically in syndication at [2] and [3], and I reckon it might
be worth seeing if there are any ideas that can be pinched from XTM
[4].

Cheers,
Danny.

[1] http://www.w3.org/2001/sw/Europe/reports/thes/1.0/guide/
[2] http://matt.blogs.it/specs/ENT/1.0/
[3] http://www.xfml.org/spec/1.0.html
[4] http://www.topicmaps.org/xtm/


-- 

http://dannyayers.com

Reply via email to