Re: qsf

On 10 Jan 2007, Darxus-5+gNnNaL/BD/[email protected] wrote:
> I got curious abou the qsf that Anthony Campbell mentioned:
> 
> ----- Forwarded message from Andrew Wood -----
> 
> > I'm considering changing to using qsf.  It looks identical to spamprobe
> > which I've been using for years (multi-word token bayesian filter written
> > in C).  How is qsf better?
> 
> I'm not familiar with spamprobe enough to give a point-by-point comparison.
> QSF has some additional features built into its tokeniser so it can detect
> common spammer tricks and treat them as tokens in their own right (it
> creates pseudo-tokens such as "comments-in-HTML" and
> "gibberish-consonants"). It's also small, fast, and accurate. Other than
> that, it's a reasonably standard Bayesian-like filter.
> 
> - Andrew

>From my limited experience so far, I'd say run qsf and spamprobe
together. Currently I'm using spamprobe to catch whatever qsf misses,
which is about 5 or 6 a day after very limited training (about 150
spams). They seem to complement each other well. I'm seeing no spams at
all and no false positives. 

Anthony

-- 
Anthony Campbell - ac-MeA/[email protected] 
Microsoft-free zone - Using Linux Gnu-Debian
http://www.acampbell.org.uk (blog, book reviews, 
on-line books and sceptical articles)


-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV