Robert Menschel <[EMAIL PROTECTED]> writes:

> The trick will be to gain the hoped for benefit without weakening the
> rules.

I could see some benefit in moving from manual phrase rules to
computer-generated rules (using the SA paragraphing technique and so on)
to get better bang from the multi-word phrases.  I think the main
utility of these is:

  - feeding autolearning with universally high probability phrases
  - two word matching (whereas bayes does one-word matching)

Daniel

-- 
Daniel Quinlan                     anti-spam (SpamAssassin), Linux,
http://www.pathname.com/~quinlan/    and open source consulting

Reply via email to