Hi,

Can someone give me some hints on how to improve the spam deteection rate 
?

we use spamassassin 3.1.0 with dcc, razor, bayes and the following rules 
via rulesdujour:

bash-2.05# cat /etc/rulesdujour/config
[ "${TRUSTED_RULESETS}" ] || \
        TRUSTED_RULESETS="TRIPWIRE ANTIDRUG RANDOMVAL BOGUSVIRUS 
BLACKLIST_URI \
                        SARE_REDIRECT_POST300 SARE_EVILNUMBERS0 
SARE_EVILNUMBERS1 SARE_EVILNUMBERS2 \
                        SARE_BAYES_POISON_NXM SARE_HTML SARE_HEADER 
SARE_SPECIFIC SARE_ADULT SARE_BML \
                        SARE_FRAUD SARE_SPOOF SARE_RANDOM 
SARE_SPAMCOP_TOP200 SARE_OEM SARE_GENLSUBJ \
                        SARE_UNSUB SARE_URI0 SARE_URI1 SARE_WHITELIST 
SARE_OBFU ";

Can we add extra rulesets ? I checked the wiki page on rulesdujour and 
these are the only "official" rules apparently. Can I use other rules with 
rules du jour ?

BTW, we have other (older I guess) rules as well in our rules directory, 
this is the current list:

10_misc.cf                     25_textcat.cf 70_sare_uri1.cf
20_advance_fee.cf              25_uribl.cf 70_sare_whitelist.cf
20_anti_ratware.cf             30_text_de.cf 70_sc_top200.cf
20_body_tests.cf               30_text_fr.cf 72_sare_bml_post25x.cf
20_compensate.cf               30_text_it.cf 72_sare_redirect_post3.0.0.cf
20_dnsbl_tests.cf              30_text_nl.cf 88_FVGT_body.cf
20_drugs.cf                    30_text_pl.cf 88_FVGT_headers.cf
20_fake_helo_tests.cf          30_text_pt_br.cf 88_FVGT_rawbody.cf
20_head_tests.cf               50_scores.cf 88_FVGT_subject.cf
20_html_tests.cf               60_awl.cf 88_FVGT_uri.cf
20_meta_tests.cf               60_whitelist.cf 99_FVGT_Tripwire.cf
20_net_tests.cf                60_whitelist_spf.cf 
99_sare_fraud_post25x.cf
20_phrases.cf                  60_whitelist_subject.cf        RulesDuJour
20_porn.cf                     70_sare_adult.cf               antidrug.cf
20_ratware.cf                  70_sare_bayes_poison_nxm.cf    backhair.cf
20_uri_tests.cf                70_sare_evilnum0.cf blacklist-uri.cf
23_bayes.cf                    70_sare_evilnum1.cf bogus-virus-warnings.cf
25_accessdb.cf                 70_sare_evilnum2.cf chickenpox.cf
25_antivirus.cf                70_sare_genlsubj.cf            languages
25_body_tests_es.cf            70_sare_header.cf              mangled.cf
25_body_tests_pl.cf            70_sare_html.cf                random.cf
25_dcc.cf                      70_sare_obfu.cf                rules.121205
25_domainkeys.cf               70_sare_oem.cf                 sober_p.cf
25_hashcash.cf                 70_sare_random.cf              triplets.txt
25_pyzor.cf                    70_sare_specific.cf            tripwire.cf
25_razor2.cf                   70_sare_spoof.cf               weeds_2.cf
25_replace.cf                  70_sare_unsub.cf
25_spf.cf                      70_sare_uri0.cf


For the moment we have on average +/- 4.5% of mails t hat score between 5 
and 9 points (we tag these mails currently, but don't drop them). Have 
other people any data on this ?

We did use pyzor before, but it caused too much trouble (lingering 
processes etc.) so we dropped it.

Any other suggestions ?


regards,
tom.





---------------------------------------------------------------------------
Tom Van Overbeke - ABSI Service Delivery Coordinator
email: [EMAIL PROTECTED]
Tel: +32 2 333 40 00 - Fax: +32 2 332 34 69
website: http://www.absi.be
---------------------------------------------------------------------------

Reply via email to