[SAtalk] chi squared combining

[Matt Tolton]

Hi List, I realize I asked this question a bit ago but nobody seemed to know anything so I'm asking again to double check:

 

What is chi squared combining and how does it affect bayesian filtering?

 

Quoting from the docs:

 

bayes_use_chi2_combining (default: 0)

Should the Bayesian classifier use chi-squared combining, instead of Robinson/Graham-style naive Bayesian combining? Chi-squared produces more 'extreme' output results, but may be more resistant to changes in corpus size etc.

 

This isn't very clear to me...can anyone give a more in-depth explanation?

 

Thankyou.

 

Matt Tolton
Network Administrator
JDI Technologies, Inc.
Ph: 818-936-7000 x124