folksonomy

[phpwiki]
Came across [a nice analysis of connetea bookmarking software|http://www.dlib.org/dlib/april05/lund/04lund.html] off a link at
del.icio.us. [Connetea|http://www.connotea.org/] seems like the right
bookmarking tool to play with
in myVocs. The [code is open|http://www.connotea.org/code] and they are
following del.icio.us ideas but
extending them in useful ways. This wouldn't elliminate del.icio.us use
but would give us a way of seemless identity integration with a
bookmarking tool.

I also took a look at the [facebook.com|http://facebook.com] site
referenced in [jim's blog|http://arch.doit.wisc.edu/jim/index.php/2005/10/04/internet2-collaboration-tools-phone-call-4-october-2005/].
This
is one of the best examples of how a [vo-core like myVocs|http://myvocs.org] should look
and work. It's got excellent methods of connecting groups to groups. It
builds excitement for the user.
Granted, this could also act as an identity provider, but could
just as easily leverage an existing idp infrastructure or be the "local
account" idp for the vo system environement.

[phpwiki]
Here's a site (http://dbacl.sourceforge.net/spam_chess-1.html) where someone uses baysian filters to play check (good/ham moves versus bad/spam) moves. This approach might help realize my idea to improve folksonomy with basian filtering (http://lab.ac.uab.edu/node/edit/1245). It really makes sense since what having many people tag the same data does is pretty much the same as way a bays filter automates. worth looking into.

[phpwiki]
this idea has been rolling around in my head for a while and the thoughts
on tagging seem to make it an even more powerful option.

bayesian filters for spam seem to be a good way to train a filter to
answer a simple yes/no question about a given piece of content. "yes this
content belongs in catagory x" or "no this content does not belong in
catagory x". Normally the catagory is "inbox". After a while of training
the filter should get pretty good at identifying terminology patterns that
associate something with catagory x or not. if you think of each person's
inbox as a distinct catagory, then there are obvious millions of distinct