1Mail::SpamAssassin::PluUgsienr::CAounMttaoriLileb:au:rtSnepTdahmrPAeessrshlaoslDsdoi(cn3u:)m:ePnltuagtiino:n:AutoLearnThreshold(3)
2
3
4
6 Mail::SpamAssassin::Plugin::AutoLearnThreshold - threshold-based
7 discriminator for Bayes auto-learning
8
10 loadplugin Mail::SpamAssassin::Plugin::AutoLearnThreshold
11
13 This plugin implements the threshold-based auto-learning discriminator
14 for SpamAssassin's Bayes subsystem. Auto-learning is a mechanism
15 whereby high-scoring mails (or low-scoring mails, for non-spam) are fed
16 into its learning systems without user intervention, during scanning.
17
18 Note that certain tests are ignored when determining whether a message
19 should be trained upon:
20
21 • rules with tflags set to 'learn' (the Bayesian rules)
22
23 • rules with tflags set to 'userconf' (user configuration)
24
25 • rules with tflags set to 'noautolearn'
26
27 Also note that auto-learning occurs using scores from either scoreset 0
28 or 1, depending on what scoreset is used during message check. It is
29 likely that the message check and auto-learn scores will be different.
30
32 The following configuration settings are used to control auto-learning:
33
34 bayes_auto_learn_threshold_nonspam n.nn (default: 0.1)
35 The score threshold below which a mail has to score, to be fed into
36 SpamAssassin's learning systems automatically as a non-spam
37 message.
38
39 bayes_auto_learn_threshold_spam n.nn (default: 12.0)
40 The score threshold above which a mail has to score, to be fed into
41 SpamAssassin's learning systems automatically as a spam message.
42
43 Note: SpamAssassin requires at least 3 points from the header, and
44 3 points from the body to auto-learn as spam. Therefore, the
45 minimum working value for this option is 6.
46
47 If the test option autolearn_force is set, the minimum value will
48 remain at 6 points but there is no requirement that the points come
49 from body and header rules. This option is useful for autolearning
50 with rules that are considered to be extremely safe indicators of
51 the spaminess of a message.
52
53 bayes_auto_learn_on_error (0 | 1) (default: 0)
54 With "bayes_auto_learn_on_error" off, autolearning will be
55 performed even if bayes classifier already agrees with the new
56 classification (i.e. yielded BAYES_00 for what we are now trying
57 to teach it as ham, or yielded BAYES_99 for spam). This is a
58 traditional setting, the default was chosen to retain backward
59 compatibility.
60
61 With "bayes_auto_learn_on_error" turned on, autolearning will be
62 performed only when a bayes classifier had a different opinion from
63 what the autolearner is now trying to teach it (i.e. it made an
64 error in judgement). This strategy may or may not produce better
65 future classifications, but usually works very well, while also
66 preventing unnecessary overlearning and slows down database growth.
67
68
69
70perl v5.32.1 Mail:2:0S2p1a-m0A4s-s1a4ssin::Plugin::AutoLearnThreshold(3)