in reply to Re: Bayesian Filtering for Spam
in thread Bayesian Filtering for Spam

regarding phrasing you could use a WPE to convert basic anonymous word tokens into phrase tokens.

did this for a recruitment company and it worked well.

Advantage is that a WPE does not need compilatio ala Yapp and can be managed via a web interface by a non techie.

Disadvantage is that it can be slow. Algorithm we developed for client was very fast but Oracle (SQL) centric so would not be a good fit for other RDBMS.

The client also used statistical methods (as discussed here) but they saw a WPE as a major plus point.

For our spam we block incoming ip address and obviously faked addresses - old hat theye days but works very well.

Jacqui