Duncan Findlay writes:

> I think so. Rules designed to catch spam, scored negatively, even if
> they occur more frequently in non-spam than spam, are NOT good
> indicators of spam. They are merely bad/false indicators of spam, and
> the regexp's should be changes to make them better spam indicators.

I agree. The first thing I do when a new SA release comes out is to add
scores to my local.cf file to cancel out any of those negative scores. I
just set most of the scores to zero except for some rules (like
HTML_WITH_BGCOLOR) that still work as excellent spam indicators for me.

There are still a couple of really wacky scores in the GA scores - for
example, the PORN_8 rule practically whitelists anything with the words
'videoz', 'mp3z', or 'warez' with a score of -4.2. While I agree this isn't
a good spam indicator, it's certainly not a good whitelist rule either.

Here's the scores from my current local.cf file - the original GA scores are
listed after the # sign on each line and are hopefully being treated as
comments.

# way out there scores
score PORN_8    0   #                      -4.248
score TRACKER_ID    0   #                  -4.215

# scores that shouldn't be negative, probably broken rules
score ALL_CAPS_SUBJECT  0   #           -0.274
score BE_AMAZED         0   #           -0.260
score GAPPY_TEXT        0   #           -1.237
score HTML_WITH_BGCOLOR 3.0   #           -0.546
score JAVASCRIPT_URI    0   #           -1.607
score LINES_OF_YELLING_3    0   #       -1.518
score NO_EXPERIENCE         0   #       -1.063
score NO_QS_ASKED           0   #       -0.773
score OPPORTUNITY           0   #       -1.010
score RATWARE               0   #       -0.703
score REAL_THING            0   #       -0.148
score RELAYING_FRAME        2.0   #       -0.584
score SLIGHTLY_UNSAFE_JAVASCRIPT        1.0   #  -0.794
score SUPERLONG_LINE                    0   #  -0.374
score SUBJ_ENDS_IN_Q_MARK               0   #  -0.050
score SUSPICIOUS_RECIPS                 0   #  -0.016
score TO_BE_REMOVED_REPLY               0   #  -2.150
score TO_UNSUB_REPLY                    0   #  -1.996
score WEB_BUGS                          2.0   #  -0.823
score X_MSMAIL_PRIORITY_HIGH            0   #  -1.356

--
michael moncur   mgm at starlingtech.com   http://www.starlingtech.com/
"You live and learn.  At any rate, you live."           -- Douglas Adams


_______________________________________________________________

Don't miss the 2002 Sprint PCS Application Developer's Conference
August 25-28 in Las Vegas -- http://devcon.sprintpcs.com/adp/index.cfm

_______________________________________________
Spamassassin-talk mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/spamassassin-talk

Reply via email to