You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@spamassassin.apache.org by bu...@bugzilla.spamassassin.org on 2007/01/09 21:41:15 UTC

[Bug 5287] New: Make auto-learning header and body thresholds configurable

http://issues.apache.org/SpamAssassin/show_bug.cgi?id=5287

           Summary: Make auto-learning header and body thresholds
                    configurable
           Product: Spamassassin
           Version: 3.1.7
          Platform: All
        OS/Version: All
            Status: NEW
          Severity: enhancement
          Priority: P5
         Component: Learner
        AssignedTo: dev@spamassassin.apache.org
        ReportedBy: nradov@axolotl.com


The Bayes auto learning thresholds can be configured using 
bayes_auto_learn_threshold_spam and bayes_auto_learn_threshold_nonspam as 
defined in AutoLearnThreshold.pm. But that file has a hard-coded minimum limit 
of 3 points in each of the body and header before a message is automatically 
learned as spam, even if the total value is above 
bayes_auto_learn_threshold_spam. What we are seeing is that most spam is 
properly identified due to the rules we have implemented, but all of those 
rules mostly only affect the header score. The body score ends up being near 0, 
so automatic learning doesn't take place.

I propose that the required_body_points and required_head_points values be made 
configurable. That way they could be tuned to work well for particular local 
configurations.



------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.

[Bug 5287] Make auto-learning header and body thresholds configurable

Posted by bu...@bugzilla.spamassassin.org.
http://issues.apache.org/SpamAssassin/show_bug.cgi?id=5287





------- Additional Comments From nradov@axolotl.com  2007-01-09 12:55 -------
Yes it would be easy to write a new autolearn discriminator plugin, but it 
would be even easier to make those two values configurable in the current 
plugin. Is there any down side to doing that? Writing a whole new plugin would 
seem like a lot of extra duplicate code just to add one minor new feature. Can 
we consider reopening this bug?



------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.

[Bug 5287] Make auto-learning header and body thresholds configurable

Posted by bu...@bugzilla.spamassassin.org.
http://issues.apache.org/SpamAssassin/show_bug.cgi?id=5287


jm@jmason.org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |WORKSFORME




------- Additional Comments From jm@jmason.org  2007-01-09 12:46 -------
this can be accomplished very easily by implementing a new autolearn
discriminator plugin to replace AutoLearnThreshold.



------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.