Read in alt.spam, posting & fups changed. I don't read in neural-nets
posted to alt.spam & comp.ai.neural-nets f/ups alt.spam
Alan McCoy wrote:
> There are numerous solutions for applying bayesian filtering on mail
> traffic, but I have yet to see usenet software using this
> technology. Is there such a thing? I'd appreciate an open source
> solution so I can hack on it, as I know it would most probably some
> manual tweaking as it's more difficult to filter usenet reliably, when
> compared to mail.
Recently there was some promotional palaver here about
CRM114, which is
a Bayesian derivative where CRM stands for Controllable Regex Mutilator,
and where the derivation expands the Bayesianess by 'sparse binary
polynomial hashing' or a bayesian chain rule or a Markovian
discrimination.
Without getting into the 'sophistication' of the Bayesian evolution,
some are moving forward with those approaches and applying them in
various ways. One of the applications is that of Dave Slusher to a news
agent application.
You can read some about CRM114 and a number of links here
http://crm114.sourceforge.net/
You can read some about the Bayesian Markovian evolution in this .pdf of
the slides from a spam conference
http://crm114.sourceforge.net/Plateau99.pdf
You can read some about the application of CRM114 to a newsreader in a
linux environment by Dave Slusher here
http://www.sff.net/people/dave_slusher/crmnewsfilter.htp "Dave Slusher
is building a Usenet News filter using the CRM114 capabilities. He's
playing with it on Usenet, news stories on Google, TV show listings, all
that cool stuff. Check it out at: The CRM114 News Filter"
If you download the Slusher tar.gz file, you will find its contents
linux based, of course, whereas much of the stuff at the sourceforge
site is also win based.
--
Mike Easter
