http://issues.apache.org/SpamAssassin/show_bug.cgi?id=4706





------- Additional Comments From [EMAIL PROTECTED]  2008-01-16 17:18 -------
I separated out the three parts of HG_HORMONE into three separate rules and also
tried three variations of the part that tests for H.G.H. Can someone who is more
familiar with interpreting the scores in deciding what rules are good please
tell me what these results mean? This is from today's nightly run. There were
1508644 spams and 143199 hams. To save space I changed entries like "0.0838 
1265 of 1508644 messages" into "0.084(1265)" rounding all decimls to two
significant places, and I removed the msecs and scores.

Spam%(num)   Ham&(num)  S/O  Rank
0.084(1265) 0.083(119) 0.50  0.62  HG_HORMONE (__HG_HORMONE same stats)
0.041(615)  0.061(88)  0.40  0.57  /\bhuman growth hormone\b/i
0.021(316)  0.035(50)  0.38  0.55  /\bHGH\b/
0.028(426)  0.0035(6)  0.89  0.62  /\bH.G.H.\b/i
0.00(0)     0.0035(6)  0.000 0.49  /\bH\.G\.H\b/
0.00(0)     0.0035(6)  0.000 0.49  /\bH\.G\.H\b/i

As far as I can tell there is no basis for testing for H.G.H because H\.G\.H has
no spam hits, and H.G.H. hits are probably coincidental. It also looks like HGH
is not good enough, or am I wrong on my interpretation of the numbers for that 
one?
 



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

Reply via email to