effectivity with local tests only [auf Virenüberprüft]

[Hans Moser <ha...@ofd-sth.niedersachsen.de>]

Hi!

My environment is a bit "weired".
I get all my Mails from my ISP's relay. So I disabled AWL.
I have no direct internet access. I can use HTTP over a proxy cascade
and SMTP with the relay host. No external DNS.

So I'm stuck with static rules and bayes. I trained SA with 1600 spam
mails and 220 ham mails.
After that I took one of the spams and invoke

# spamassassin --test-mode < spammail

[...]

Inhaltsanalyse im Detail:   (3.5 Punkte, 5.0 benötigt)

Pkte Regelname              Beschreibung
---- ----------------------
--------------------------------------------------
   0.0 HTML_MESSAGE           BODY: Nachricht enthält HTML
   3.5 BAYES_99               BODY: Spamwahrscheinlichkeit nach
Bayes-Test: 99-100%
                              [score: 1.0000]
   0.0 MIME_HTML_ONLY         BODY: MIME-Nachricht besteht nur aus HTML

Bayes ist nearly 100% certain that it is spam. It is spam, but the score
ist only 3,5.

Can you advice my something to adjust to get spam to the right score?

Set score of bayes up?
Lower score for spam?

Razor, Pyzor and DCC need direct internet access. If there is no other
way, I'll try to work this out with my ISP, but this will be hard work...


Hans