attribution, context and qualifiers

[Mark Nottingham <mnot@mnot.net>]

One of the things that annoys me about many feeds is that there isn't
a clean separation between the metadata for the channel source and
the channel aggregator.

For example, last time I looked, Meerkat takes over the channel
<titles> of its feeds, completely obliterating the actual name of the
feed. Similarly, I've seen a lot of "brought to you by..." messages
in feed titles recently. The same goes for the images that show a
message about the aggregator, rather than the channel.

To me, one of the core principles of RSS is getting all of the
metadata sorted out so that it's easily identifiable; this doesn't
help. This bleeds somewhat into the date and other DC qualifiers
issue as well. Having flexible buckets like "title" and "date" to put
metadata into RSS is good, but we need to be able to contextualise
them in a readily understandable way.

Any thoughts on how to do this? An XML-intuitive way would be to
stick an optional attribute onto elements which contextualises them;
you could then define an extensible vocabulary which allows those
who are interested to tell the difference between a <title> which is
the actual resource title, vs. the title that someone making a
comment about it has made (for instance).

I think RDF would conceptualise this as context; however, there
aren't any good ways to express context in RDF/XML yet, AFAIK. DC
defines qualifiers, but there isn't yet a syntax for expressing
qualifiers in RDF/XML or XML.

I'd like to see an intuitive, easy-to-use way to introduce
context/qualifiers into RSS which encourages the establishment and
reuse of some well-known vocabularies (like attribution, different
meanings of date, etc.). Ideally, the machine should be able to tell
that the image has been supplied by the software that generated the
channel by default, and doesn't have anything to do with the channel.
Likewise, it should be able to know the difference between the date
that the target resource was created vs. the date that the statements
(rss <item>) were made about the target resource.

Cheers,


-- 
Mark Nottingham
http://www.mnot.net/