On Tue, 16 Nov 2004 17:53:36 -0500, Bob Wyman <[EMAIL PROTECTED]> wrote:
I think the idea has a *lot* of potential. The information that needs to be carried is of a pretty generic type: (name, value) so this should make a very good use case for Atom extensibility - on topic. The syntax seems fine for Atom, but it would be nice* to be able to use the same extension syntax for RSS 1.0 and 2.0 as well. Unfortunately RDF/XML balks on this structure: <item rdf:about="http://example.org/sdf";> <ps:metric ps:scheme="ps:LinkRank">27</ps:metric> </item> "item" is the subject, "ps:metric" is the property, and object is...[*crunch*] Alternatives that should work for RSS 1.0, 2.0 and Atom (assuming appropriate namespacing) include: <item> <linkRank>27</linkRank> <classification>politics<classification> </item> <item> <metric rdf:parseType="Resource"> <name>LinkRank</name> <value>27</value> </metric> </item> <item> <metric> <Metric> <name>LinkRank</name> <value>27</value> </Metric> </metric> </item> <item> <metrics rdf:parseType="Resource"> <linkRank>27</linkRank> <classification>politics<classification> </metrics> </item> + plenty of others... > Also, if you think this makes sense, could you suggest some metrics that you > would like to see inserted into entries by in-channel processors like > PubSub.com? Hmm, most of what could be inserted in an individual feed that would be available to a client anyway, though things like update stats might be good for client auto-optimization. What I've been wanting to play with for a long while are similarity measures - how similar is this to the posts from Tim Bray say, allowing "more like this" or (not referring to Tim, 'course) "less like this"...and a dirty-word rating?... yep, all that wacky machine learning stuff that's been around for years waiting for an application... Cheers, Danny. * nice, in that at least it would allow easier XSLT, at most it would highlight modelling errors. -- http://dannyayers.com
