On 9 Oct 2006 at 11:20, Micheal Espinola Jr wrote:

> Paul,
> 
> I'm curious to know if I have explained myself well enough for you to 
> understand my reasoning.  I value your (as well as Fritz's, and anyone 
> else's) opinion about this.

Well, I've been away from the list for a couple of days, but I see you've had 
some response!

I have been using several different Bayesian-type filters for some time.  I am 
currently running 
another filter in parallel to ASSP for comparison/evaluation.  It continues to 
amaze me how 
accurate and how adaptive this method is - ASSP's 2-word tokens particularly.  
Even spam containing 
just images are caught.  It is often quite a surprise to see what is 
significant in the corpus or 
what tokens were used in classification - and impossible to second-guess.

I have a site were internal mail is very different to external, but I don't 
find the Bayesian 
classifier is phased by this at all.  (In the same way that a spammer quoting 
Dickens or poetry in 
an attempt to upset the filtering is still caught.)  You have a different 
experience.

Now there are as many different 'mail mixes' as there are people on this list, 
so I won't pretend 
that my issues are the same as yours, nor vice-versa.  One of the many 
excellent points about ASSP 
is it's versatility and adaptability - you've made it perform the way you 
wanted it to.

Paul


-------------------------------------------------------------------------
Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job easier
Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
_______________________________________________
Assp-user mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/assp-user

Reply via email to