>Corpus-Norm: 1.04099504248725 this is nearly perfect - it leads in to a corpus confidence of 1.000
>Corrected-NotSpamFiles: 8 this shows that assp had 8 false positives over all the time - where is the problem ? >SpamWords/File: 106 >Hamwords/File: 567 both values are a bit less (I expect : 180 and 650) - you may try to increase 'MaxBytes' by 2000 (max 8000 !) notice: after you changed this value, delete the normfile and run the rebuildspamdb twice - ignore the result (rebuildrun.txt) of the first run >So if I see this correctly it looks at if my database is spam heavy. Only a little bit (4%) - every value less than 10% is acceptable. Every rebuildspamdb task (except the first) will try to get 0% - corpus norm = 1.000 >I am dealing with false positives correct? No - not really - because there are too less corrected false positives (8)! You may have false positives, but you don't 'deal' with! Thomas Von: Jay <[email protected]> An: [email protected] Datum: 03.09.2015 20:54 Betreff: Re: [Assp-user] Bayesian settings So I think I am narrowing down my issue here with ASSP and I am just looking for some verification here. So looking at the email interface under Rebuild Spamdb I opened up normfile and found some interesting information. Corpus-Norm: 1.04099504248725 Corrected-SpamFiles: 235 Corrected-NotSpamFiles: 8 Spamlog-Files: 15458 NotSpamlog-Files: 2604 SpamWords/File: 106 Hamwords/File: 567 Spamwords: 1952635 Hamwords: 1875739 So if I see this correctly it looks at if my database is spam heavy. So in turn this would cause the issue I am dealing with false positives correct? On 9/1/2015 5:39 PM, Jay wrote: > So I have a question about the settings for 'DoBayesian'. I am still > having issues where my users are complaining about legitimate emails > getting blocked. I would analyze some of those emails in the email > interface mail analyzer and I get the result of spam probability of > 1.000000. Looking at the word pairs I see a lot of common words in there > listed as bad word matches. What I did last week was take a previous > spamdb file we had built before we switched to MySQL as the database > backend and in had ASSP import that file by using the file extension of > RPL. That worked fine. I tested 1 email before and after, I got better > results afterwards. > > Before database replace the test email was flagged as spam. After > database replace the test email was listed as good. So there was a > definite improvement. There's still a lingering an issue somewhere in > our database flagging so many emails as spam when they are legitimate. > My question is this, for the DoBayesian setting what are the > repercussions of setting this to score? Currently we are set to block, > which from my understanding will block emails if it is determined to be > spam. Will scoring do the same after reaching a certain score? I don't > want to just switch the settings and find that my users are all of the > sudden getting bombarded with spam. That would just aggravate them and > not something I want to happen. > > Here's the current settings we are running: > > Bayesian & HMM Options > > DoBayesian = block > DoHMM = disabled > BayesAfterHMM = empty > BayesWL = off > BayesNP = off > BayesLocal = off > noBayesian = empty > maxByesValues = 60 > baysProbability = 0.6 > BayesConf = 0 > baysConfidenceHalfScore = on > > ASSP Worker/DB/Regex status = healthy > > > > > > ------------------------------------------------------------------------------ > _______________________________________________ > Assp-user mailing list > [email protected] > https://lists.sourceforge.net/lists/listinfo/assp-user > > > > ----- > No virus found in this message. > Checked by AVG - www.avg.com > Version: 2015.0.6086 / Virus Database: 4409/10562 - Release Date: 09/02/15 > > ------------------------------------------------------------------------------ Monitor Your Dynamic Infrastructure at Any Scale With Datadog! Get real-time metrics from all of your servers, apps and tools in one place. SourceForge users - Click here to start your Free Trial of Datadog now! http://pubads.g.doubleclick.net/gampad/clk?id=241902991&iu=/4140 _______________________________________________ Assp-user mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/assp-user DISCLAIMER: ******************************************************* This email and any files transmitted with it may be confidential, legally privileged and protected in law and are intended solely for the use of the individual to whom it is addressed. This email was multiple times scanned for viruses. There should be no known virus in this email! ******************************************************* ------------------------------------------------------------------------------ _______________________________________________ Assp-user mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/assp-user
