On Fri, 9 Jul 2010 16:44:57 +0200, Norman Uittenbogaart <norm...@gmail.com> wrote: > Hello Stevan, > > Results are way better at the moment, thank you! > > nor...@*.nl: > TP True Positives: 774 > TN True Negatives: 2109 > FP False Positives: 3 > FN False Negatives: 290 > SC Spam Corpusfed: 0 > NC Nonspam Corpusfed: 0 > TL Training Left: 388 > SHR Spam Hit Rate 72.74% > HSR Ham Strike Rate: 0.14% > PPV Positive predictive value: 99.61% > OCA Overall Accuracy: 90.77% > Well... OSB is way, way better then WORD or just CHAIN. Anyway... I see you are still in the training phase (388 trainings left). After that value is down to zero you will see again a increase of accuracy.
-- Kind Regards from Switzerland, Stevan Bajić > > On Mon, Jun 28, 2010 at 9:34 AM, Norman Uittenbogaart <nor...@tbu.nl> > wrote: >> Hello Stevan put the new config in hope it helps, thanks ! :) >> I'm using shared,managed for several reasons, it is in front of another >> spam >> filter so lots of spam isn't coming through that I can train, now I can >> also >> train those mails. >> I have users who don't understand forwarding email for spam training and >> working with quarantines. >> >> >> On Sun, Jun 27, 2010 at 12:26 PM, Stevan Bajić <ste...@bajic.ch> wrote: >>> >>> On Fri, 25 Jun 2010 17:39:13 +0200 >>> Norman Uittenbogaart <nor...@tbu.nl> wrote: >>> >>> > Sounds great thanks ! >>> > >>> I attached your dspam.conf that you sent me with some changes. The main >>> change is that I switched the tokenizer to OSB and added a bunch of >>> IgnoreHeader lines. And I changed the PurgeXXXX lines since you are >>> probably >>> purging with the help of the SQL purge script and the purge days are >>> specified there and not in dspam.conf. >>> >>> You probably should empty the old tokens form your DSPAM installation by >>> running, after you have exchanged your old dspam.conf with the new one: >>> TRUNCATE `dspam_signature_data`; >>> TRUNCATE `dspam_stats`; >>> TRUNCATE `dspam_token_data`; >>> >>> You wrote that you use "shared,managed" for your user. Might I ask why? >>> >>> -- >>> Kind Regards from Switzerland, >>> >>> Stevan Bajić >>> >>> >>> > On Fri, Jun 25, 2010 at 4:24 PM, Stevan Bajić <ste...@bajic.ch> >>> > wrote: >>> > >>> > > I am currently out of office working on a project. As soon as I get >>> > > home I >>> > > will quickly look over your dspam.conf and send you a bunch of >>> > > suggestions. >>> > > Okay? >>> > > >>> > > >>> > > >>> > > >>> > > >>> > > -- >>> > > Kind Regards from Switzerland, >>> > > >>> > > Stevan Bajić >>> > > >>> > > On Fri, 25 Jun 2010 16:15:50 +0200, Norman Uittenbogaart < >>> > > norm...@gmail.com> wrote: >>> > > >>> > > Ok thats not a big problem as this is not really working out..... >>> > > Which one should I use ? >>> > > >>> > > >>> > > On Fri, Jun 25, 2010 at 3:58 PM, Stevan Bajić <ste...@bajic.ch> >>> > > wrote: >>> > > >>> > >> Hallo Norman, >>> > >> >>> > >> >>> > >> >>> > >> getting higher accuracy would require to start from the beginning >>> > >> since >>> > >> the tokenizer you use is not that intelligent and using anything >>> > >> other more >>> > >> intelligent would require you to start from the beginning with the >>> > >> learning. >>> > >> >>> > >> >>> > >> >>> > >> >>> > >> >>> > >> -- >>> > >> Kind Regards from Switzerland, >>> > >> >>> > >> Stevan Bajić >>> > >> >>> > >> On Fri, 25 Jun 2010 15:18:37 +0200, Norman Uittenbogaart < >>> > >> norm...@gmail.com> wrote: >>> > >> >>> > >> Hi Stevan, >>> > >> >>> > >> Any suggestions to getting the accuracy up would be great ! >>> > >> My user is shared,managed in the groups file for the domains used >>> > >> now.. >>> > >> >>> > >> nor...@t*.nl: >>> > >> TP True Positives: 856 >>> > >> TN True Negatives: 6650 >>> > >> FP False Positives: 6 >>> > >> FN False Negatives: 2097 >>> > >> SC Spam Corpusfed: 0 >>> > >> NC Nonspam Corpusfed: 0 >>> > >> TL Training Left: 0 >>> > >> SHR Spam Hit Rate 28.99% >>> > >> HSR Ham Strike Rate: 0.09% >>> > >> PPV Positive predictive value: 99.30% >>> > >> OCA Overall Accuracy: 78.11% >>> > >> >>> > >> >>> > >> On Fri, Jun 25, 2010 at 3:02 PM, Stevan Bajić <ste...@bajic.ch> >>> > >> wrote: >>> > >> >>> > >>> Hallo Norman, >>> > >>> >>> > >>> >>> > >>> >>> > >>> you asked "what is wrong". Right? Nothing is wrong with your DSPAM >>> > >>> installation. There are things you could do better but nothing is >>> > >>> really, >>> > >>> really wrong. >>> > >>> >>> > >>> >>> > >>> >>> > >>> Could you post the output of "dspam_stats -H ". >>> > >>> >>> > >>> >>> > >>> >>> > >>> >>> > >>> >>> > >>> -- >>> > >>> Kind Regards from Switzerland, >>> > >>> >>> > >>> Stevan Bajić >>> > >>> >>> > >>> >>> > >>> >>> > >>> >>> > >>> >>> > >>> On Fri, 25 Jun 2010 14:51:30 +0200, Norman Uittenbogaart < >>> > >>> norm...@gmail.com> wrote: >>> > >>> >>> > >>> Hello Stevan, >>> > >>> >>> > >>> Herewith enclosed my config. >>> > >>> >>> > >>> Kind regards, >>> > >>> Norman >>> > >>> >>> > >>> On Fri, Jun 25, 2010 at 2:25 PM, Stevan Bajić <ste...@bajic.ch> >>> > >>> wrote: >>> > >>> >>> > >>>> Hello, >>> > >>>> >>> > >>>> >>> > >>>> >>> > >>>> the statistic alone is not enough information for telling why >>> > >>>> your >>> > >>>> accuracy is down. You need to post your dspam.conf. >>> > >>>> >>> > >>>> >>> > >>>> >>> > >>>> >>> > >>>> >>> > >>>> -- >>> > >>>> Kind Regards from Switzerland, >>> > >>>> >>> > >>>> Stevan Bajić >>> > >>>> >>> > >>>> >>> > >>>> >>> > >>>> >>> > >>>> >>> > >>>> On Fri, 25 Jun 2010 11:29:06 +0200, Norman Uittenbogaart < >>> > >>>> norm...@gmail.com> wrote: >>> > >>>> >>> > >>>> I have been running and training dspam for while now, but it is >>> > >>>> not >>> > >>>> at >>> > >>>> all effective at the moment. >>> > >>>> What could be wrong ? >>> > >>>> >>> > >>>> SPAM messages Good messages Since last reset 2083 missed 6 >>> > >>>> missed 856 >>> > >>>> caught 6558 delivered 29.126% caught 0.091% missed Total >>> > >>>> processed by >>> > >>>> filter 2097 missed 6 missed 856 caught 6559 delivered From >>> > >>>> corpus >>> > >>>> 0 >>> > >>>> fed 0 fed >>> > >>>> Metric Calculated as Overall accuracy (since last reset) >>> > >>>> *78.017%* (SPAM >>> > >>>> messages caught + Good messages delivered) / Total number of >>> > >>>> messages Spam >>> > >>>> identification (since last reset) *29.126%* (Spam catch rate >>> > >>>> only) >>> > >>>> Spam >>> > >>>> ratio (of total processed) *31.025%* Total SPAM messages (both >>> > >>>> caught & >>> > >>>> missed) / Total number of message >>> > >>>> >>> > >>>> >>> > >>>> >>> > >>>> >>> > >>>> ------------------------------------------------------------------------------ >>> > >>>> ThinkGeek and WIRED's GeekDad team up for the Ultimate >>> > >>>> GeekDad Father's Day Giveaway. ONE MASSIVE PRIZE to the >>> > >>>> lucky parental unit. See the prize list and enter to win: >>> > >>>> http://p.sf.net/sfu/thinkgeek-promo >>> > >>>> _______________________________________________ >>> > >>>> Dspam-user mailing list >>> > >>>> Dspam-user@lists.sourceforge.net >>> > >>>> https://lists.sourceforge.net/lists/listinfo/dspam-user >>> > >>>> >>> > >>>> >>> > >>> >>> > >>> >>> > >>> >>> > >>> ------------------------------------------------------------------------------ >>> > >>> ThinkGeek and WIRED's GeekDad team up for the Ultimate >>> > >>> GeekDad Father's Day Giveaway. ONE MASSIVE PRIZE to the >>> > >>> lucky parental unit. See the prize list and enter to win: >>> > >>> http://p.sf.net/sfu/thinkgeek-promo >>> > >>> _______________________________________________ >>> > >>> Dspam-user mailing list >>> > >>> Dspam-user@lists.sourceforge.net >>> > >>> https://lists.sourceforge.net/lists/listinfo/dspam-user >>> > >>> >>> > >>> >>> > >> >>> > >> >>> > >> >>> > >> >>> > >> ------------------------------------------------------------------------------ >>> > >> ThinkGeek and WIRED's GeekDad team up for the Ultimate >>> > >> GeekDad Father's Day Giveaway. ONE MASSIVE PRIZE to the >>> > >> lucky parental unit. See the prize list and enter to win: >>> > >> http://p.sf.net/sfu/thinkgeek-promo >>> > >> _______________________________________________ >>> > >> Dspam-user mailing list >>> > >> Dspam-user@lists.sourceforge.net >>> > >> https://lists.sourceforge.net/lists/listinfo/dspam-user >>> > >> >>> > >> >>> > > >>> > > >>> > > >>> > > ------------------------------------------------------------------------------ >>> > > ThinkGeek and WIRED's GeekDad team up for the Ultimate >>> > > GeekDad Father's Day Giveaway. ONE MASSIVE PRIZE to the >>> > > lucky parental unit. See the prize list and enter to win: >>> > > http://p.sf.net/sfu/thinkgeek-promo >>> > > _______________________________________________ >>> > > Dspam-user mailing list >>> > > Dspam-user@lists.sourceforge.net >>> > > https://lists.sourceforge.net/lists/listinfo/dspam-user >>> > > >>> > > >>> >>> >>> ------------------------------------------------------------------------------ >>> This SF.net email is sponsored by Sprint >>> What will you do first with EVO, the first 4G phone? >>> Visit sprint.com/first -- http://p.sf.net/sfu/sprint-com-first >>> _______________________________________________ >>> Dspam-user mailing list >>> Dspam-user@lists.sourceforge.net >>> https://lists.sourceforge.net/lists/listinfo/dspam-user >>> >> >> ------------------------------------------------------------------------------ This SF.net email is sponsored by Sprint What will you do first with EVO, the first 4G phone? Visit sprint.com/first -- http://p.sf.net/sfu/sprint-com-first _______________________________________________ Dspam-user mailing list Dspam-user@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/dspam-user
