With Bayesian enabled.. wouldn't it have a higher hit count? I didn't see that in your list of test matches...
----- Original Message ----- From: "Michael Moncur" <[EMAIL PROTECTED]> To: "Justin Mason" <[EMAIL PROTECTED]> Cc: <[EMAIL PROTECTED]> Sent: Friday, December 06, 2002 7:48 AM Subject: RE: [SAtalk] 2.50CVS false positives <big snip> 4. For possible use in meta rules, their messages always match CTYPE_JUST_HTML and WEB_BUGS, usually SUPERLONG_LINE and JAVASCRIPT, and not much else. I ran one of these through 2.50CVS and it does a bit better with the new HTML percentage test, among other things: SPAM: ---- Start SpamAssassin results SPAM: 8.40 hits, 5 required; SPAM: * 1.8 -- BODY: Message is 90-100% HTML tags SPAM: * 1.2 -- BODY: Javascript to open a new window SPAM: * 1.0 -- BODY: HTML has unbalanced "html" tags SPAM: * 0.8 -- BODY: Javascript to move windows around SPAM: * 0.2 -- BODY: Image tag with an ID code to identify you SPAM: * 0.2 -- BODY: JavaScript code SPAM: * 0.0 -- BODY: T_HTML_P2_90_100 SPAM: * 0.0 -- BODY: T_HTML_IMAGE_AREA01 SPAM: * 0.0 -- BODY: T_HTML_MESSAGE SPAM: * 0.0 -- BODY: T_HTML_P1_80_100 SPAM: * 0.0 -- BODY: T_HTML_TAG_EXISTS_CENTER SPAM: * 0.0 -- BODY: T_HTML_SHOUTING1 SPAM: * 0.0 -- BODY: T_HTML_WIN_FOCUS SPAM: * 0.0 -- BODY: T_HTML_CONSEC_IMGS04 SPAM: * 0.0 -- BODY: T_HTML_NUM_IMGS07 SPAM: * 2.7 -- Listed in DCC, see http://rhyolite.com/anti-spam/dcc/ SPAM: * 0.5 -- HTML-only mail, with no text version SPAM: SPAM: ---- End of SpamAssassin results -- Michael Moncur mgm at starlingtech.com http://www.starlingtech.com/ "Only the shallow know themselves." --Oscar Wilde ------------------------------------------------------- This sf.net email is sponsored by:ThinkGeek Welcome to geek heaven. http://thinkgeek.com/sf _______________________________________________ Spamassassin-talk mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/spamassassin-talk
