-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Larry Gilson writes:
>Hi Bob,
>
>Sorry for the long delay in my response.  I have taken a little break.
>Thanks for running the rules through masscheck against your corpus.  I have
>no where near the corpus that you do and find the testing methodology in
>your first and second run, and results very interesting.  Thanks again!

Robert, BTW --

were you saying previously that you couldn't get hit-frequencies working?
Could you remind me what was up there?   I might be able to help.

- --j.

>
>
>> -----Original Message-----
>> From: Robert Menschel
>>
>> Tuesday, November 25, 2003, 10:12:27 PM, I wrote:
>> 
>> LG>> Attached is a custom rule file.  It has been working rather well 
>> LG>> and I will be increasing the score from 0.5 to 1.0.  The cf file 
>> LG>> also has some rules looking for words obfuscated by pipes.  They 
>> LG>> have been working well also.
>> 
>> RM> FYI, My masscheck results with your rules (run against my corpus 
>> RM> of 58,857 emails). Final number on each line is what I would 
>> RM> initially score them based on these hits (per my algorithm posted 
>> RM> at http://www.exit0.us/index.php/RM_RuleScoring -- most sites 
>> RM> should probably score these lower, and I would probably want to 
>> RM> do a 2-pass or 3-pass GA on these to refine the scores myself).
>> 
>> Ran a second mass-check pass, assigning those scores 
>> (repeated below), and with those scores,
>> * I had no FPs based on these rules -- highest score in any ham was 
>>   less than 6.0 out of 9.0
>> * These rules by themselves, with these scores, correctly flagged 263 
>>   as spam with scores of 9.0 or higher.
>> 
>> I'm guessing that others could take my scores, adjust for 
>> threshold and other rules, and use these rules productively.  
>> If you score spam at 5.0, then you'd want to multiply each of 
>> my scores by 5/9 (or maybe a little less, on the theory that 
>> other rules should hit the spam also).
>> 
>> My own hesitation rests with that "and other rules" part ... 
>> If this ruleset by itself can bring a ham to (say) 4.0 of 
>> 9.0, and that ham matches 5.0 worth of rules without this 
>> ruleset, it'll then be wrongly flagged as spam.
>> 
>> I am just about at the point where I can do a mass-check 
>> including all of my custom rules, but I haven't yet figured 
>> out how to incorporate all of the distributed rules in that 
>> mass-check.
>> 
>> I'm supplying my ruleset in testdir/spamassassin/user_prefs, 
>> and pointing to that directory using the -c parameter to 
>> mass-check. If I were to copy all of the 2.60 *.cf rules 
>> files into that testdir/spamassassin directory, would that 
>> activate the default rules for the mass-check test?
>> 
>> Bob Menschel
>> 
>> RM> MY_RBDY_PDS_1P3   -- 375s /  22h -- 1.163
>> RM> MY_RBDY_PDS_1P4   -- 365s /   5h -- 1.608
>> RM> MY_RBDY_PDS_1P5   -- 210s /   3h -- 1.700
>> RM> MY_RBDY_PDS_1P6   -- 165s /   2h -- 0.550
>> RM> MY_RBDY_PDS_1P7   --  88s /   0h -- 1.880
>> RM> MY_RBDY_PDS_1P8   -- 121s /   4h -- 1.302
>> RM> MY_RBDY_PDS_2P2   -- 168s /  14h -- 1.112
>> RM> MY_RBDY_PDS_2P3   -- 105s /  45h -- 0.228
>> RM> MY_RBDY_PDS_2P4   -- 311s /  14h -- 1.207
>> RM> MY_RBDY_PDS_2P5   --  56s /   7h -- 0.700
>> RM> MY_RBDY_PDS_2P6   -- 161s /   8h -- 1.179
>> RM> MY_RBDY_PDS_2P7   --  89s /   5h -- 1.148
>> RM> MY_RBDY_PDS_2P8   --   4s /   5h -- 0.067
>> RM> MY_RBDY_PDS_3P1   -- 200s /  15h -- 1.125
>> RM> MY_RBDY_PDS_3P2   -- 173s /  25h -- 6.654
>> RM> MY_RBDY_PDS_3P3   -- 179s /  58h -- 0.303
>> RM> MY_RBDY_PDS_3P4   --  74s /  15h -- 0.463
>> RM> MY_RBDY_PDS_3P5   -- 195s /  12h -- 1.150
>> RM> MY_RBDY_PDS_3P6   --  43s /   5h -- 0.717
>> RM> MY_RBDY_PDS_3P7   --   3s /   5h -- 0.050
>> RM> MY_RBDY_PDS_3P8   --  42s /  49h -- 0.084
>> RM> MY_RBDY_PDS_4P1   -- 285s /  32h -- 0.864
>> RM> MY_RBDY_PDS_4P2   -- 417s /  21h -- 1.190
>> RM> MY_RBDY_PDS_4P3   -- 259s /  82h -- 0.312
>> RM> MY_RBDY_PDS_4P4   -- 160s /  26h -- 0.593
>> RM> MY_RBDY_PDS_4P5   --  56s /  17h -- 0.311
>> RM> MY_RBDY_PDS_4P6   --   7s /   0h -- 0.700
>> RM> MY_RBDY_PDS_4P7   --   3s /  12h -- 0.023
>> RM> MY_RBDY_PDS_4P8   --   2s /   0h -- 0.200
>> RM> MY_RBDY_PDS_5P1   --  84s /  21h -- 0.382
>> RM> MY_RBDY_PDS_5P3   --  99s / 464h -- 0.021
>> RM> MY_RBDY_PDS_5P5   --  81s /  12h -- 0.623
>> RM> MY_RBDY_PDS_6P6   --  99s / 464h -- 0.021
>> RM> MY_HDR_PDS_1P5    -- 140s /   0h -- 2.400
>> RM> MY_HDR_PDS_2P1    -- 244s /   3h -- 1.610
>> RM> MY_HDR_PDS_2P4    -- 176s /  13h -- 1.126
>> RM> MY_HDR_PDS_3P2    -- 308s /   9h -- 1.308
>> RM> MY_HDR_PDS_3P3    -- 607s / 528h -- 0.115
>> RM> MY_HDR_PDS_3P5    -- 108s /   0h -- 2.080
>> RM> MY_HDR_PDS_3P8    --  73s /   0h -- 1.730
>> RM> MY_HDR_PDS_4P3    -- 481s / 519h -- 0.093
>> RM> MY_HDR_PDS_4P4    -- 114s /  13h -- 0.814
>> RM> MY_HDR_PDS_4P5    --  82s /   0h -- 1.820
>> RM> MY_HDR_PDS_5P1    -- 171s /   0h -- 2.710
>> RM> MY_HDR_PDS_6P1    -- 159s /   0h -- 2.590
>> RM> MY_HDR_PDS_6P2    -- 122s /   9h -- 1.122
>> RM> MY_BDY_PIPE_S233S --  17s /   0h -- 1.170
>> RM> MY_BDY_PIPE_S23S  --  35s /   0h -- 1.350
>> RM> MY_BDY_PIPE_S23C  --  17s /   0h -- 1.170
>> RM> MY_BDY_PIPE_S24S  --  42s /   0h -- 1.420
>> RM> MY_BDY_PIPE_S34P  --   0s /   0h -- 0.100
>> RM> MY_HDR_PIPE_S233S --   0s /   0h -- 0.100
>> RM> MY_HDR_PIPE_S23S  --   0s /   0h -- 0.100
>> RM> MY_HDR_PIPE_S23C  --   0s /   0h -- 0.100
>> RM> MY_HDR_PIPE_S24S  --   0s /   0h -- 0.100
>> RM> MY_HDR_PIPE_S34P  --   0s /   0h -- 0.100
>
>
>
>-------------------------------------------------------
>This SF.net email is sponsored by: SF.net Giveback Program.
>Does SourceForge.net help you be more productive?  Does it
>help you create better code?  SHARE THE LOVE, and help us help
>YOU!  Click Here: http://sourceforge.net/donate/
>_______________________________________________
>Spamassassin-talk mailing list
>[EMAIL PROTECTED]
>https://lists.sourceforge.net/lists/listinfo/spamassassin-talk
>
>
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.2 (GNU/Linux)
Comment: Exmh CVS

iD8DBQE/y4Y0QTcbUG5Y7woRAvG4AKDtDH/1W2pkIDjslAI08jm7b9sLAQCgh9Lw
ZEPja01WDgiGXwNuVXfYI/I=
=FfM7
-----END PGP SIGNATURE-----



-------------------------------------------------------
This SF.net email is sponsored by: SF.net Giveback Program.
Does SourceForge.net help you be more productive?  Does it
help you create better code?  SHARE THE LOVE, and help us help
YOU!  Click Here: http://sourceforge.net/donate/
_______________________________________________
Spamassassin-talk mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/spamassassin-talk

Reply via email to