Nobody? Nothing?
Update on poor User 3:
TP True Positives: 0
TN True Negatives: 421
FP False Positives: 1
FN False Negatives: 119
SC Spam Corpusfed: 0
NC Nonspam Corpusfed: 0
TL Training Left: 2078
SHR Spam Hit Rate 0.00%
HSR Ham Strike Rate: 0.24%
PPV Positive predictive value: 0.00%
OCA Overall Accuracy: 77.82%
Shouldn’t dspam be catching at least *some* spam by now?
Can anyone suggest what I can analyze, or what tests I can do to try to
understand why this user is getting zero spam protection from dspam while other
users are getting good results with identical settings?
Thanks,
pjv
On Feb 26, 2014, at 8:17 PM, Plexus LLC <[email protected]> wrote:
> Hi,
>
> I’m having a problem getting spam marked for a particular user. Below is the
> output of spam_stats -H for three users. User 3 is the obvious problem. User
> 2 is my own personal email account upon which dspam is working great. User 1
> is another user who is having good results.
>
> User 3 is newer to the server, but has been diligently retraining spam for
> several weeks now (I have dovecot-antispam plugin set up) and dspam has not
> yet marked one spam message as anything other than (I)nnocent.
>
> All of the users on the system have the exact same configuration - there are
> no individual level configs or preferences in effect. I am obviously not
> using any groups - each user has their own individual dictionary. Training
> mode is set to TEFT.
>
> I’m afraid I am a total noob with troubleshooting dspam. I’m a sysadmin and I
> set up the mail server, but since I haven’t had any trouble till now, I don’t
> really know where to start in figuring out the problem and in this case my
> usually formidable google-fu has failed me.
>
> What can I do to try to diagnose why dspam is forsaking User 3?
>
> Thanks for any and all help,
> pjv
>
>
> User 1:
> TP True Positives: 5599
> TN True Negatives: 2007
> FP False Positives: 152
> FN False Negatives: 69
> SC Spam Corpusfed: 0
> NC Nonspam Corpusfed: 0
> TL Training Left: 341
> SHR Spam Hit Rate 98.78%
> HSR Ham Strike Rate: 7.04%
> PPV Positive predictive value: 97.36%
> OCA Overall Accuracy: 97.18%
>
> User 2:
> TP True Positives: 263
> TN True Negatives: 2015
> FP False Positives: 1
> FN False Negatives: 18
> SC Spam Corpusfed: 0
> NC Nonspam Corpusfed: 0
> TL Training Left: 484
> SHR Spam Hit Rate 93.59%
> HSR Ham Strike Rate: 0.05%
> PPV Positive predictive value: 99.62%
> OCA Overall Accuracy: 99.17%
>
> User 3:
> TP True Positives: 0
> TN True Negatives: 377
> FP False Positives: 1
> FN False Negatives: 103
> SC Spam Corpusfed: 0
> NC Nonspam Corpusfed: 0
> TL Training Left: 2122
> SHR Spam Hit Rate 0.00%
> HSR Ham Strike Rate: 0.26%
> PPV Positive predictive value: 0.00%
> OCA Overall Accuracy: 78.38%
> ------------------------------------------------------------------------------
------------------------------------------------------------------------------
Subversion Kills Productivity. Get off Subversion & Make the Move to Perforce.
With Perforce, you get hassle-free workflows. Merge that actually works.
Faster operations. Version large binaries. Built-in WAN optimization and the
freedom to use Git, Perforce or both. Make the move to Perforce.
http://pubads.g.doubleclick.net/gampad/clk?id=122218951&iu=/4140/ostg.clktrk
_______________________________________________
Dspam-user mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/dspam-user
