[SURBL-Discuss] Question about scoring in SA3

Jeff Chan jeffc at surbl.org
Wed May 18 15:53:45 CEST 2005

On Wednesday, May 18, 2005, 5:52:30 AM, Kevin McGrail wrote:
> I've been very pleased with SURBL in SA3 and I'd like to increase the
> scores.  However, I don't understand how the default scores like this work:

> rules/50_scores.cf:score URIBL_AB_SURBL 0 2.007 0 0.417
> rules/50_scores.cf:score URIBL_OB_SURBL 0 1.996 0 3.213
> rules/50_scores.cf:score URIBL_PH_SURBL 0 0.839 0 2.000
> rules/50_scores.cf:score URIBL_SC_SURBL 0 3.897 0 4.263
> rules/50_scores.cf:score URIBL_WS_SURBL 0 0.539 0 1.462

> I feel SA is being too conservative with the resource that SURBL provides.
> Can anyone give me their recommendations for my local configuration file for
> replacement scores that will be more effective?

> And since I couldn't find it referenced, can anyone tell me what the four
> numbers after the score mean?

Here's a reference for the four values once mentioned on the
SpamAssassin Users list by Theo in response to my own similar

> $ perldoc Mail::SpamAssassin::Conf
> [...]
>    If four valid scores are listed, then the score that is used
>    depends on how SpamAssassin is being used. The first score is used
>    when both Bayes and network tests are disabled (score set 0). The
>    second score is used when Bayes is disabled, but network tests are
>    enabled (score set 1). The third score is used when Bayes is
>    enabled and network tests are disabled (score set 2). The fourth
>    score is used when Bayes is enabled and network tests are enabled
>    (score set 3).

You're certainly free to increase or decrease any scores you
like.  In case it's of interest the scores are set using a type
of neural net called a perceptron in order to optimize them
against the SpamAssassin test corpora.  These default scores tend
to work well, but everyone's own local corpora of spam and ham
may be different.

I'm sure the folks on the SpamAssasin list may have more thoughts
about this.


Jeff C.
Don't harm innocent bystanders.

More information about the Discuss mailing list