Re: Subject/Categorization Data

[Brian McCallister <mccallister@xxxxxxxxxxxxxxxxxxx>]

Hmm, DublinCore would work fine for me =)

What needs to happen for this to be blessed into the spec?

-Brian

On Feb 5, 2004, at 5:41 PM, Nick Chalko wrote:

> Current use seems to be
>
> <dc:subject 
> xmlns:dc="http://purl.org/dc/elements/1.1/";>SelfReference</dc:subject>
>
> If I would do something like this
> <subject source="http://www.tnrdlib.bc.ca/dewey.html"; 
> key="507">Education, research, related topics<subject>
>
> With source and key optional.  And all of it completely arbitrary.
>
>
>
>
>
>
>
> Brian McCallister wrote:
>
> > Subject should be of the Content type (allowing the various mime type 
> > declarations and and mode declarations).
> >
> > Because their *are* so many ways to categorize things, and 
> > categorization systems are rapidly changing, I don't think a hard 
> > system is usefully specifiable -- build for the flexibility Adam 
> > talks about =)
> >
> > <subject>
> >     Dogs in Space
> > </subject>
> >
> > <subject mode="escaped">
> >     Dogs in Space
> > </subject>
> >
> > <subject mode="xml">
> >     <pets>
> >         <space>Dogs</space>
> >     </pets>
> > </subject>
> >
> > This means that a client will need to understand the scheme being 
> > used by its upstream provider if it wants to make use of the 
> > categorization -- which is FINE. Let separate standards for arbitrary 
> > categorization emerge and let Atom worry about syndication rather 
> > than reinventing another dewy decimal system.
> >
> > -Brian