> > Thunderbird has a Bayesian junk/spam detector that will
> > automatically learn to detect spam based on what you mark
> > as spam.
>
> except that it's dumb as a post...even after extensive training...
>
>
> > Thunderbird is pretty good if your mail volume isn't HUGE.
>
> ah, that might be the problem.

Totally disagree... My mail volumes (maybe not large to some) are any
where between 500-5000 per day (peaking at large spam times to 14000)
and I'm perfectly happy with Thunderbirds spam filters and
performance.

Bayesian filters has been proven to be the most proficient (as long as
there is sufficient training data) for eliminating/reducing spam. I'm
currently running about 0.001% incorrect classified emails / day.

popfile is still a bayesian filtering system and procmail is simply
rule based. Isn't there more upkeep with rule based filters since you
have to update the rule set each time the spammer gets smarter / tries
something new ??

The best thing is that now it can correctly identify new 'types' of
spam since it does not relay on simplistic rule base filtering.

Just my $0.02 worth...

/paul



> > It is very much better than anything else I have used, even
> > Postini service from my dialup ISP.
>
> i use popfile. once trained it catches everything. clever html, other
> languages, etc.
>
> i've been telling friends for a few years now: if they're complaining
> about spam, their not using popfile.
>
> --
> http://home.cogeco.ca/~tsummerfelt1
> telnet://ventedspleen.dyndns.org
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [EMAIL PROTECTED]
> For additional commands, e-mail: [EMAIL PROTECTED]
>
>

---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to