Re: BayesIt or Bayes Filter?
Hello I have not re-installed it either. So I don't know BayesIt suffers from the same problem that affected Bayes Filter ... at least earlier in the beta series. I find this thread really interesting, because BayesIT was so buggy in earlier releases that it drove me (an many others) to the Bayes Filter. IMHO, Bayes Filter is FAR faster and more stable than BayesIT ever was for me. The plugin issue in the last beta series has been resolved in Bayes Filter 2.0.3 so I am back using this, and these are the metrics (reset stats after installing 2.0.3 earlier this week):- HamMails: 18191 SpamMails: 4063 detected Ham: 2087 (99.86%) detected Spam: 94 (93.62%) FALSE Ham detected: 6 FALSE Spam detected: 3 .. I think this is pretty good. Can other systems do better? -- Regards Graham -- Notice: This e-mail message, together with any attachments, contains information of Merck Co., Inc. (One Merck Drive, Whitehouse Station, New Jersey, USA 08889), and/or its affiliates (which may be known outside the United States as Merck Frosst, Merck Sharp Dohme or MSD and in Japan, as Banyu) that may be confidential, proprietary copyrighted and/or legally privileged. It is intended solely for the use of the individual or entity named on this message. If you are not the intended recipient, and have received this message in error, please notify us immediately by reply e-mail and then delete it from your system. -- Current beta is 3.5.26 | 'Using TBBETA' information: http://www.silverstones.com/thebat/TBUDLInfo.html IMPORTANT: To register as a Beta tester, use this link first - http://www.ritlabs.com/en/partners/testers/
Re: BayesIt or Bayes Filter?
Hello Graham, Wednesday, June 8, 2005, 11:18:41 AM, you wrote: FG HamMails: 18191 FG SpamMails: 4063 FG detected Ham: 2087 (99.86%) FG detected Spam: 94 (93.62%) FG FALSE Ham detected: 6 FG FALSE Spam detected: 3 FG .. I think this is pretty good. Can other systems do better? K9 report: Since ma apr 26 2004 02:32:21 (408 days) Server Check Total number of emails processed 110,697 271/day Number of Good emails processed 104,084 94.03% Number of Spam emails processed6,613 5.97% Percentage of emails that matched whitelist rules 0.00% Percentage of emails that matched blacklist rules 0.38% Number of emails re-classified to Good 7 Number of emails re-classified to Spam 664 Percent misidentified as Spam (false positives)0.01% Percent misidentified as Good (false negatives)0.60% Accuracy 99.39% Personal notes: The 7 emails had to be re-classified to Good because I personally had (by clicking the wrong button) re-classiefied them to Spam before. ± 90% of the 664 re-classified to Spam had been caught by TB! filters Most of those are caught by the rule that looks for text cotaining no full stop (., that is) while the sender is not in the address book. Another type of messages that K9 had problems with are those that contain normal text that doesn't have anything to do with the attached gif that tries to sell _i_g_a, c_a_i_ and ch_a_ software (I'm not witing it in full: one might never know if this message gets caught by an anti-spam filter :-) ) Most of those get caught by making a filter look for =?utf-8?B? or =?utf-8?q? while the address of the sender isn't in the addressbook. -- Best Wishes, Mark using The Bat! Version 3.5.25 with MyMacros 1.11 Useless Macro Collection 2.1.1121 rc8 zOmbie's Macros Version 0.7 Windows 2000 Professional/5.0 build 2195 Service Pack 4 (0 days 2:38:44) on Uno AMD Duron Current beta is 3.5.26 | 'Using TBBETA' information: http://www.silverstones.com/thebat/TBUDLInfo.html IMPORTANT: To register as a Beta tester, use this link first - http://www.ritlabs.com/en/partners/testers/
Re[2]: BayesIt or Bayes Filter?
Hello Simon, Wednesday, June 8, 2005, 3:59:55 AM, you wrote: SF Spambayes was more resource intensive on my machine than SF Popfile. Within 10 minutes of using it 200MB of Virtual RAM was SF committed along with 100MB of RAM. The program relentlessly SF thrashed the Hard Disc when asking it to load the configuration SF page - not good. If you are looking for an alternative, a lot of folks that have tried K9 really like it. It is very popular with folks on the Gibson Research forums. The developer hangs out there and offers great support. -- Best regards, MikeDmailto:[EMAIL PROTECTED] Using The Bat! v3.5.25 on Windows XP 5.1 Build 2600 Service Pack 2 Current beta is 3.5.26 | 'Using TBBETA' information: http://www.silverstones.com/thebat/TBUDLInfo.html IMPORTANT: To register as a Beta tester, use this link first - http://www.ritlabs.com/en/partners/testers/
Re[2]: BayesIt or Bayes Filter?
Hello Graham, Wednesday, June 8, 2005, 4:18:41 AM, you wrote: FG Hello I have not re-installed it either. So I don't know BayesIt suffers from the same problem that affected Bayes Filter ... at least earlier in the beta series. FG I find this thread really interesting, because BayesIT was so buggy in FG earlier releases that it drove me (an many others) to the Bayes FG Filter. FG IMHO, Bayes Filter is FAR faster and more stable than BayesIT ever was FG for me. The plugin issue in the last beta series has been resolved in FG Bayes Filter 2.0.3 so I am back using this, and these are the FG metrics (reset stats after installing 2.0.3 earlier this week):- I guess I have been lucky. BayesIt (0.8.0) has been working well for me for quite some time. Now I don't get quite that high on my spam classification, but I have never gotten a false positive which is the one thing I do not want ever. I am willing to have a few spam messages to deal with if it means I don't have to worry about missing something important. -- Best regards, MikeDmailto:[EMAIL PROTECTED] Using The Bat! v3.5.25 on Windows XP 5.1 Build 2600 Service Pack 2 Current beta is 3.5.26 | 'Using TBBETA' information: http://www.silverstones.com/thebat/TBUDLInfo.html IMPORTANT: To register as a Beta tester, use this link first - http://www.ritlabs.com/en/partners/testers/
Re[3]: BayesIt or Bayes Filter?
Wednesday, June 8, 2005, 7:01:07 AM, (Internet Time - @542) you wrote: Hello Mike, M3 If you are looking for an alternative, a lot of folks that have tried M3 K9 really like it. It is very popular with folks on the Gibson M3 Research forums. The developer hangs out there and offers great M3 support. I have to agree with you on K9. I have been using it with The Bat! for about a year now and it really does a nice job. Uses very little memory and works great with all my accounts. I have it set up to filter all my e-mail accounts before The Bat! filters them. So far it is setting at about 98.8% with the spam messages. And the best thing about K9 is it is Freeware! If anyone wants to try it out, you can find it here: http://keir.net Hope this helps... -- Best regards, Michael mailto:[EMAIL PROTECTED] Using The Bat! Version 3.5.26 On Windows XP SP2 Current beta is 3.5.26 | 'Using TBBETA' information: http://www.silverstones.com/thebat/TBUDLInfo.html IMPORTANT: To register as a Beta tester, use this link first - http://www.ritlabs.com/en/partners/testers/
Re[2]: BayesIt or Bayes Filter?
Wednesday, June 8, 2005, 4:18:41 AM, (Internet Time - @429) you wrote: Hello Graham, FG I find this thread really interesting, because BayesIT was so buggy in FG earlier releases that it drove me (an many others) to the Bayes FG Filter. FG IMHO, Bayes Filter is FAR faster and more stable than BayesIT ever was FG for me. The plugin issue in the last beta series has been resolved in FG Bayes Filter 2.0.3 so I am back using this, and these are the FG metrics (reset stats after installing 2.0.3 earlier this week):- FG HamMails: 18191 FG SpamMails: 4063 FG detected Ham: 2087 (99.86%) FG detected Spam: 94 (93.62%) FG FALSE Ham detected: 6 FG FALSE Spam detected: 3 FG .. I think this is pretty good. Can other systems do better? Here's my stats from K9: Column 1: Since Fri Nov 19 2004 04:56:37 PM (201 days) Column 2: Since Fri Nov 19 2004 04:56:37 PM (201 days) Total number of emails processed 24,86924,869 124/day 124/day Number of Good emails processed7,402 7,402 29.76%29.76% Number of Spam emails processed 17,46717,467 70.24%70.24% Percentage of emails that matched whitelist rules 0.00% 0.00% Percentage of emails that matched blacklist rules 0.00% 0.00% Number of emails re-classified to Good 148 148 Number of emails re-classified to Spam 126 126 Percent misidentified as Spam (false positives)0.60% 0.60% Percent misidentified as Good (false negatives)0.51% 0.51% 98.90%98.90% -- Best regards, Michael mailto:[EMAIL PROTECTED] Using The Bat! Version 3.5.26 On Windows XP SP2 Current beta is 3.5.26 | 'Using TBBETA' information: http://www.silverstones.com/thebat/TBUDLInfo.html IMPORTANT: To register as a Beta tester, use this link first - http://www.ritlabs.com/en/partners/testers/
Re[2]: BayesIt or Bayes Filter?
Hello Simon, On Tuesday, June 07, 2005 at 22:48:59 GMT +0100 (which was 23:48:59 where I live), Simon Fincham wrote and made these valuable points on the subject of BayesIt or Bayes Filter?: Hi Dick, On 07 June 2005 at 20:18:09GMT +0200, Dick Hoogendoorn sent an E-Mail mid:[EMAIL PROTECTED] on the subject of BayesIt or Bayes Filter?: Dick I gave up on both products and installed Spambayes a few Dick months ago. Never let me down and I only trained it for about Dick one week and since then there is hardly a spammail that will not Dick be classified as such, in the worst case it's classified as Dick Unsure and ends up in my Unsure folder. I've read with interest your experience with 'Spambayes'. I previously had good results with 'Popfile' but found it to be slow and resource intensive, so in search of an integrated solution went back to 'Bayesit'. After training 'Bayesit' with a large base it still refuses to capture 'obvious' Spam messages - so, I'm going to try 'Spambayes' and see what it can do. Thanks. Simon, I've used POPFile for a while but for whatever reason, it took ages to read the emails from 5 pop3 accounts while Spambayes does the same in a few seconds. I even played with POPFile's option of using multiple pop sessions to speed things up but it did not help. -- Best regards, Dick ___ Those of you who think you know everything are very annoying to those of us who do. Using The Bat! v3.5.25 on Windows XP 5.1 Build 2600 Service Pack 2 Privacy is your right...only if you defend it: http://www.metropipe.net/landing.cgi?id=cloggy Current beta is 3.5.26 | 'Using TBBETA' information: http://www.silverstones.com/thebat/TBUDLInfo.html IMPORTANT: To register as a Beta tester, use this link first - http://www.ritlabs.com/en/partners/testers/
Re[3]: BayesIt or Bayes Filter?
Hello Dick, I have never had a problem with the Bayes filter. BayesIt is just to buggy. -- Best regards, rmorris.r TheBat!3.5.25 AntiSpam=Bayes Filter Plugin v2.0.3 mailto:[EMAIL PROTECTED] Current beta is 3.5.26 | 'Using TBBETA' information: http://www.silverstones.com/thebat/TBUDLInfo.html IMPORTANT: To register as a Beta tester, use this link first - http://www.ritlabs.com/en/partners/testers/
Re: Re[3]: BayesIt or Bayes Filter?
I have to agree with you on K9. I have been using it with The Bat! for about a year now and it really does a nice job. Uses very little memory and works great with all my accounts. I have it set up to filter all my e-mail accounts before The Bat! filters them. So far it is setting at about 98.8% with the spam messages. Another POP utility. So far, the only IMAP spam filter I know of is built into Eudora. On the other hand, I'm not sure I agree with the concept of a spam filter. It searches for offending words and uses a probability algorithm, right? The problem is that there are people who can send me sexy emails, and there are also people that I buy things from. I find that a purchase confirmation will end up in Eudora's junk file almost without fail. I get pretty good results by specifying likely offending phrases (like investor alert or cialis) and having TB put them in the junk folder. Those go 100% of the time. -- Gleason UsingTheBat!v3.5.26onWindowsXP5.1Build2600 ServicePack2PrimarilyusingtheFastmail IMAPserverwhichusesCyrus. Current beta is 3.5.26 | 'Using TBBETA' information: http://www.silverstones.com/thebat/TBUDLInfo.html IMPORTANT: To register as a Beta tester, use this link first - http://www.ritlabs.com/en/partners/testers/
Re: BayesIt or Bayes Filter?
Gleason, On 08-06-2005 17:41, you [GP] wrote in mid:[EMAIL PROTECTED]: GP Another POP utility. So far, the only IMAP spam filter I know of is GP built into Eudora. SpamPal supposedly works for IMAP. -- greeting Best regards /greeting author Peter Fjelsten /author thebat version 3.5.21 Pro /thebat versionextras MyGate, AVG /extras env. ~11 POP3, 2 IMAP (MailMax 5.5) 1 IMAP (Exchange 6.5), 175K msgs. /env. os Windows XP 5.1 Build 2600 Service Pack 2 /os Current beta is 3.5.26 | 'Using TBBETA' information: http://www.silverstones.com/thebat/TBUDLInfo.html IMPORTANT: To register as a Beta tester, use this link first - http://www.ritlabs.com/en/partners/testers/
Re[5]: BayesIt or Bayes Filter?
Hello Gleason, Wednesday, June 8, 2005, 10:41:58 AM, you wrote: GP On the other hand, I'm not sure I agree with the concept of a spam GP filter. It searches for offending words and uses a probability GP algorithm, right? The problem is that there are people who can GP send me sexy emails, and there are also people that I buy things GP from. I find that a purchase confirmation will end up in Eudora's GP junk file almost without fail. Then the Bayes filter is not working correctly. It is not a word filter in the sense of the old Outlook Express mail rules. There are some great explanations of how Bayesian filters work on the interenet that are much better than anything I could type up here. You should understand that information if you want to know why bayesian filtering is such a useful tool ... at least for now. Having said that, there are a lot of little tweeks that need to be done to make a bayesian filter work well. Mine has never gotten a false positive, but that is because I have the threshold set relatively high, and so have more spam get through. GP I get pretty good results by specifying likely offending phrases GP (like investor alert or cialis) and having TB put them in the junk GP folder. Those go 100% of the time. Until they use C I A L 1 S or other manglings that you can read but slide right through your word filters. It is an ever escalating battle. We come up with new tools to protect us, they come up with new ways around the tools sigh -- Best regards, MikeDmailto:[EMAIL PROTECTED] Using The Bat! v3.5.25 on Windows XP 5.1 Build 2600 Service Pack 2 Current beta is 3.5.26 | 'Using TBBETA' information: http://www.silverstones.com/thebat/TBUDLInfo.html IMPORTANT: To register as a Beta tester, use this link first - http://www.ritlabs.com/en/partners/testers/
BayesIt or Bayes Filter?
Dear all, I give it an other try. But before installing TB! v3.5.26 with an AntiSpam Tool I like to consult this ML and see what experience you have made with either of these Tools mentioned in the Subject line. Thank you! -- best regards| Using The Bat! v3.0.1.33 www.EddieCastelli.com | on Windows 2000 5.0 Eddie | Build 2195 Service Pack 3 on Tour | Current beta is 3.5.26 | 'Using TBBETA' information: http://www.silverstones.com/thebat/TBUDLInfo.html IMPORTANT: To register as a Beta tester, use this link first - http://www.ritlabs.com/en/partners/testers/
Re: BayesIt or Bayes Filter?
On 2005-06-07 (18:09), you wrote:
I give it an other try. But before installing TB! v3.5.26 with an
AntiSpam Tool I like to consult this ML and see what experience you
have made with either of these Tools mentioned in the Subject line.
Moderately usless - both of them. I tryed them and after some time of good
work, they failed - effectiveness became lower and lower. I used them with
18 accounts with ~600 and more e-mails per day. Among those acounts was
'abuse@' type account - in this case I'd rather like to don't use
antispam tool, but in The Bat! there is no other way: all account
or nothing. This stategy is wrong on my position. But I think that
mentioned tools are good enough for home users.
Now I use Spamihilator (local proxy) and it works like charm.
Best regards, Krzysztof Kudacik
--
pb _, )\__/( ,_ Polska Strona Programu The Bat! |
/'.;`-.`{..}',-';.`\ http://thebat.pl |
/'.'; `-,`--',-' ;`.`\ The Bat! Polish Support |
\/\/^\---.\;;/.---/^\/\/ |
3.5.25 : Windows XP 5.1 Build 2600 Dodatek Service Pack 2
:krk:
Current beta is 3.5.26 | 'Using TBBETA' information:
http://www.silverstones.com/thebat/TBUDLInfo.html
IMPORTANT: To register as a Beta tester, use this link first -
http://www.ritlabs.com/en/partners/testers/
Re: BayesIt or Bayes Filter?
Hello Eddie, Tuesday, June 7, 2005, 11:09:02 AM, you wrote: EC Dear all, EC I give it an other try. But before installing TB! v3.5.26 with an EC AntiSpam Tool I like to consult this ML and see what experience you EC have made with either of these Tools mentioned in the Subject line. EC Thank you! I am still using BayesIt 0.8.0 and not having any problems. But then I have not re-installed it either. So I don't know BayesIt suffers from the same problem that affected Bayes Filter ... at least earlier in the beta series. -- Best regards, MikeDmailto:[EMAIL PROTECTED] Using The Bat! v3.5.25 on Windows XP 5.1 Build 2600 Service Pack 2 Current beta is 3.5.26 | 'Using TBBETA' information: http://www.silverstones.com/thebat/TBUDLInfo.html IMPORTANT: To register as a Beta tester, use this link first - http://www.ritlabs.com/en/partners/testers/
Re[2]: BayesIt or Bayes Filter?
Hello MikeD, On Tuesday, June 07, 2005 at 11:09:28 GMT -0500 (which was 18:09:28 where I live), MikeD (3) wrote and made these valuable points on the subject of BayesIt or Bayes Filter?: Hello Eddie, Tuesday, June 7, 2005, 11:09:02 AM, you wrote: EC Dear all, EC I give it an other try. But before installing TB! v3.5.26 with an EC AntiSpam Tool I like to consult this ML and see what experience you EC have made with either of these Tools mentioned in the Subject line. EC Thank you! I am still using BayesIt 0.8.0 and not having any problems. But then I have not re-installed it either. So I don't know BayesIt suffers from the same problem that affected Bayes Filter ... at least earlier in the beta series. I gave up on both products and installed Spambayes a few months ago. Never let me down and I only trained it for about one week and since then there is hardly a spammail that will not be classified as such, in the worst case it's classified as Unsure and ends up in my Unsure folder. -- Best regards, Dick ___ It's hard to make a comeback when you haven't been anywhere. Using The Bat! v3.5.26 on Windows XP 5.1 Build 2600 Service Pack 2 Privacy is your right...only if you defend it: http://www.metropipe.net/landing.cgi?id=cloggy Current beta is 3.5.26 | 'Using TBBETA' information: http://www.silverstones.com/thebat/TBUDLInfo.html IMPORTANT: To register as a Beta tester, use this link first - http://www.ritlabs.com/en/partners/testers/
Re: BayesIt or Bayes Filter?
On 6/7/05, Eddie Castelli [EMAIL PROTECTED] wrote: Dear all, I give it an other try. But before installing TB! v3.5.26 with an AntiSpam Tool I like to consult this ML and see what experience you have made with either of these Tools mentioned in the Subject line. Thank you! Just to save you some aggravation, don't bother if you're using IMAP. UCE protection is not implemented for IMAP. -- Kevin Current beta is 3.5.26 | 'Using TBBETA' information: http://www.silverstones.com/thebat/TBUDLInfo.html IMPORTANT: To register as a Beta tester, use this link first - http://www.ritlabs.com/en/partners/testers/
Re[3]: BayesIt or Bayes Filter?
Hello Dick, Tuesday, June 7, 2005, 1:17:28 PM, you wrote: DH I gave up on both products and installed Spambayes a few DH months ago. Never let me down and I only trained it for about one DH week and since then there is hardly a spammail that will not be DH classified as such, in the worst case it's classified as Unsure DH and ends up in my Unsure folder. The proxies are nice if they work for you. I cannot use a proxy because none of them support MSN's authentication sigh They are also just another process that wants to steal CPU cycles g -- Best regards, MikeDmailto:[EMAIL PROTECTED] Using The Bat! v3.5.25 on Windows XP 5.1 Build 2600 Service Pack 2 Current beta is 3.5.26 | 'Using TBBETA' information: http://www.silverstones.com/thebat/TBUDLInfo.html IMPORTANT: To register as a Beta tester, use this link first - http://www.ritlabs.com/en/partners/testers/
Re: BayesIt or Bayes Filter?
I give it an other try. But before installing TB! v3.5.26 with an AntiSpam Tool I like to consult this ML and see what experience you have made with either of these Tools mentioned in the Subject line. Thank you! Just to save you some aggravation, don't bother if you're using IMAP. UCE protection is not implemented for IMAP. Discovered that a while back. Yes, it is a big aggravation. Wouldn't it be great if that was included in the planned IMAP improvements? -- Gleason UsingTheBat!v3.5.26onWindowsXP5.1Build2600 ServicePack2PrimarilyusingtheFastmail IMAPserverwhichusesCyrus. Current beta is 3.5.26 | 'Using TBBETA' information: http://www.silverstones.com/thebat/TBUDLInfo.html IMPORTANT: To register as a Beta tester, use this link first - http://www.ritlabs.com/en/partners/testers/
Re: BayesIt or Bayes Filter?
On Tuesday, June 07, 2005 it appears that Simon Fincham wrote the following in regards to BayesIt or Bayes Filter?: SF [snip] After training 'Bayesit' with a large base it still refuses to capture SF 'obvious' Spam messages - so, I'm going to try 'Spambayes' and see what it can SF do. [/snip] Simon. That's been my frustrating experience, too I'm going to give spambayes a shot as well. -- Jan Rifkinson Ridgefield, CT USA TB!3.5.26, BayesIt! 0.8.1, Windows 2000, Service Pack 4 ICQ 41116329 Current beta is 3.5.26 | 'Using TBBETA' information: http://www.silverstones.com/thebat/TBUDLInfo.html IMPORTANT: To register as a Beta tester, use this link first - http://www.ritlabs.com/en/partners/testers/
