http://bugzilla.spamassassin.org/show_bug.cgi?id=3285





------- Additional Comments From [EMAIL PROTECTED]  2004-04-21 11:34 -------
Subject: Re:  SUSPICIOUS_RECIPS triggers on "[EMAIL PROTECTED]" <[EMAIL 
PROTECTED]>

> The results are really crappy for me:
>
> old:  0.037   0.0446   0.0000    1.000   0.92    2.90  FAKED_UNDISC_RECIPS
> new:  0.007   0.0075   0.0000    1.000   1.00    2.90  FAKED_UNDISC_RECIPS

This rule didn't change with that check-in, so the change is something
else.

The nightly corpus run results are somewhat more promising:

SORTED_RECIPS before:

  0.187   0.3854   0.0000    1.000   0.92    4.30  SORTED_RECIPS:jm
  0.309   0.4615   0.0000    1.000   0.96    4.30  SORTED_RECIPS:lan
  2.772   5.4836   0.0000    1.000   0.93    4.30  SORTED_RECIPS:parkerm
  0.017   0.0200   0.0134    0.600   0.20    4.30  SORTED_RECIPS:quinlan
  2.930   2.9650   0.0000    1.000   0.92    4.30  SORTED_RECIPS:theo

and after:

  0.187   0.3854   0.0000    1.000   0.92    4.30  SORTED_RECIPS:jm
  0.309   0.4615   0.0000    1.000   0.96    4.30  SORTED_RECIPS:lan
  2.773   5.4902   0.0000    1.000   0.93    4.30  SORTED_RECIPS:parkerm
  0.010   0.0200   0.0000    1.000   0.93    4.30  SORTED_RECIPS:quinlan
  2.897   2.9302   0.0000    1.000   0.92    4.30  SORTED_RECIPS:theo

clearly an improvement overall

SUSPICIOUS_RECIPS before:

  0.625   1.2892   0.0000    1.000   0.92    2.63  SUSPICIOUS_RECIPS:jm
  0.356   0.5325   0.0000    1.000   0.96    2.63  SUSPICIOUS_RECIPS:lan
  3.999   7.8984   0.0129    0.998   0.93    2.63  SUSPICIOUS_RECIPS:parkerm
  0.047   0.0400   0.0534    0.428   0.07    2.63  SUSPICIOUS_RECIPS:quinlan
  0.267   0.2686   0.1683    0.615   0.21    2.63  SUSPICIOUS_RECIPS:theo
total    10.0287   0.2346

and after:

  0.560   1.1563   0.0000    1.000   0.92    2.63  SUSPICIOUS_RECIPS:jm
  0.356   0.5325   0.0000    1.000   0.96    2.63  SUSPICIOUS_RECIPS:lan
  3.980   7.8664   0.0128    0.998   0.93    2.63  SUSPICIOUS_RECIPS:parkerm
  0.030   0.0400   0.0200    0.666   0.27    2.63  SUSPICIOUS_RECIPS:quinlan
  0.257   0.2578   0.1641    0.611   0.21    2.63  SUSPICIOUS_RECIPS:theo
total     9.853    0.1969

Borderline, but a slight improvement in total S/O ratio.

VERY_SUSP_RECIPS before:

  0.019   0.0399   0.0000    1.000   0.92    2.90  VERY_SUSP_RECIPS:jm
  0.024   0.0355   0.0000    1.000   0.96    2.90  VERY_SUSP_RECIPS:lan
  0.000   0.0000   0.0000    0.500   0.12    2.90  VERY_SUSP_RECIPS:parkerm
  0.003   0.0067   0.0000    1.000   0.93    2.90  VERY_SUSP_RECIPS:quinlan
  0.006   0.0057   0.0000    1.000   0.92    2.90  VERY_SUSP_RECIPS:theo

and after:

  0.019   0.0399   0.0000    1.000   0.92    2.90  VERY_SUSP_RECIPS:jm
  0.024   0.0355   0.0000    1.000   0.96    2.90  VERY_SUSP_RECIPS:lan
  0.000   0.0000   0.0000    0.500   0.12    2.90  VERY_SUSP_RECIPS:parkerm
  0.003   0.0067   0.0000    1.000   0.93    2.90  VERY_SUSP_RECIPS:quinlan
  0.005   0.0053   0.0000    1.000   0.92    2.90  VERY_SUSP_RECIPS:theo

No real change.

Not a huge benefit, but I'd favor keeping the change since it hopefully
fixes the reported problem.  The thresholds and limits might bear some
updating, it's been a year or two since they were set.





------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

Reply via email to