On Mon, 19 Apr 2010 18:59:24 +0200 Stevan Bajić <ste...@bajic.ch> wrote:
> Begin forwarded message: > > Date: Mon, 19 Apr 2010 16:46:49 +0000 > From: "Edward P. Ross" <epr...@acrocat.com> > To: dspam-user@lists.sourceforge.net > Subject: TOE vs TEFT: New user > Date: Mon, 19 Apr 2010 18:46:34 +0200 (CEST) > User-Agent: SquirrelMail/1.4.8-5.el4.centos.8 > Message-ID: <22867.90.52.202.167.1271695594.squir...@mail.acrocatlabs.com> > > I have a user that has the following stats: > > Total processed by filter > Spam Good > 169 missed 15 missed > 821 caught 62 delivered > > Spam ratio is at 92.7% and overall accuracy is much lower than other users. > What Spam ratio? Do you mean Spam Hit Rate (SHR)? > Currently I am running everyone with TEFT, but I just read that TOE might > work better for this user. Given that they are relatively new, does > switching to TOE make sense? > I would say that TOE would not make much difference in your case with that low amount of data. At least not much (if at all). The user has less than 1000 processed messages. That is nothing. Probably you are using CHAIN as tokenizer (you almost have 1/4 FP/FN and that leads me to believe that you have one of the simpler/dull tokenizers (aka: WORD or CHAIN)). Right? > Edward > -- Kind Regards from Switzerland, Stevan Bajić ------------------------------------------------------------------------------ Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev _______________________________________________ Dspam-user mailing list Dspam-user@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/dspam-user
