[phpBB Debug] PHP Notice: in file [ROOT]/includes/session.php on line 2208: Array to string conversion
[phpBB Debug] PHP Notice: in file [ROOT]/includes/session.php on line 2208: Array to string conversion
[phpBB Debug] PHP Notice: in file [ROOT]/includes/session.php on line 2208: Array to string conversion
[phpBB Debug] PHP Notice: in file [ROOT]/includes/session.php on line 2208: Array to string conversion
[phpBB Debug] PHP Notice: in file [ROOT]/includes/session.php on line 2208: Array to string conversion
[phpBB Debug] PHP Notice: in file [ROOT]/includes/session.php on line 2208: Array to string conversion
[phpBB Debug] PHP Notice: in file [ROOT]/includes/session.php on line 2208: Array to string conversion
[phpBB Debug] PHP Notice: in file [ROOT]/includes/session.php on line 2208: Array to string conversion
[phpBB Debug] PHP Notice: in file [ROOT]/includes/session.php on line 2208: Array to string conversion
[phpBB Debug] PHP Notice: in file [ROOT]/includes/session.php on line 2208: Array to string conversion
[phpBB Debug] PHP Notice: in file [ROOT]/includes/session.php on line 2208: Array to string conversion
[phpBB Debug] PHP Notice: in file [ROOT]/includes/session.php on line 2208: Array to string conversion
[phpBB Debug] PHP Notice: in file [ROOT]/includes/session.php on line 2208: Array to string conversion
[phpBB Debug] PHP Notice: in file [ROOT]/includes/session.php on line 2208: Array to string conversion
[phpBB Debug] PHP Notice: in file [ROOT]/includes/session.php on line 2208: Array to string conversion
[phpBB Debug] PHP Notice: in file [ROOT]/includes/session.php on line 2208: Array to string conversion
[phpBB Debug] PHP Notice: in file [ROOT]/includes/session.php on line 2208: Array to string conversion
[phpBB Debug] PHP Notice: in file [ROOT]/includes/session.php on line 2208: Array to string conversion
[phpBB Debug] PHP Notice: in file [ROOT]/includes/session.php on line 2208: Array to string conversion
[phpBB Debug] PHP Notice: in file [ROOT]/includes/session.php on line 2208: Array to string conversion
[phpBB Debug] PHP Notice: in file [ROOT]/includes/session.php on line 2208: Array to string conversion
[phpBB Debug] PHP Notice: in file [ROOT]/includes/session.php on line 2208: Array to string conversion
[phpBB Debug] PHP Notice: in file [ROOT]/includes/session.php on line 2208: Array to string conversion
[phpBB Debug] PHP Warning: in file [ROOT]/includes/functions.php on line 4688: Cannot modify header information - headers already sent by (output started at [ROOT]/includes/functions.php:3823)
[phpBB Debug] PHP Warning: in file [ROOT]/includes/functions.php on line 4690: Cannot modify header information - headers already sent by (output started at [ROOT]/includes/functions.php:3823)
[phpBB Debug] PHP Warning: in file [ROOT]/includes/functions.php on line 4691: Cannot modify header information - headers already sent by (output started at [ROOT]/includes/functions.php:3823)
[phpBB Debug] PHP Warning: in file [ROOT]/includes/functions.php on line 4692: Cannot modify header information - headers already sent by (output started at [ROOT]/includes/functions.php:3823)
Poco Forums • View topic - Spam Blockers

Spam Blockers

General email topics, from anti-virus and anti-spam software to webmail and ISPs

Moderators: Eric, Tomas, robin

Spam Blockers

Postby J-Mac » Thu Dec 02, 2004 11:24 am

I can't explain it.

I had luck like yours with Pocomail's junk filters, whether using the regular non-bayesian or using the strict bayesian. Yet others are getting great results. A real mixed bag of results, if you search these forums!

Some say it may drop for a bit until you get it trained well, but I trained it - using a couple of different methods based on various users' instructions - and gave up when it kept dropping and I got tired of dealing with the spam.

Back when I was using Outlook I used iHateSpam from Sunbelt Software - about 82% was the best accuracy rate I got with a lot of false positives. With K9 I'm getting 94% right now. Some use Popmail also.
J-Mac
J-Mac
Poco Enthusiast
 
Posts: 356
Joined: Wed Jul 28, 2004 9:54 pm
Location: The Great State of Pennsylvania, in the Merry Old Land of Oz!

Postby ambkla » Thu Dec 02, 2004 1:09 pm

Spam

I have used and am still using Spamihilator (freeware) and prior to this I used K9. Both were constantly at 98% or better. This is quite a lot bearing in mind that I get 150 - 180 e-mails on a quiet day.

I chose Spamihilator because it takes only a mouse click to undo a mail incorrectly categorized as spam. In addition, it checks e-mail in the background for any number of accounts and pops-up a window when good mail arrives.

Important: You should have a Whitelist (*@domain.com) and around 4 or 5 of the available add-ons: domain filter, header filter, etc. Spami itself uses Bayesian. It can be set to link to a Mail.exe to call the mail program when good mail drops in.

I love it.

Klaus
ambkla
Poco Tourist
 
Posts: 20
Joined: Tue Nov 30, 2004 4:13 am
Location: Oregon

Postby Alycat » Fri Dec 03, 2004 11:38 am

I came home from work last night and before starting Poco I installed K9, switched off checking of emails by Avast and switched off Poco's spam filters.

I then connected and downloaded the emails since the morning, there were the usual 60 to 70, K9 "missed" most of them, went through them all and marked then as good/spam. Last night and into today I have received about another 80 emails, all but 2 spam have been marked as spam. There has been no good emails marked as spam.

I am very hard to please...but WOW!

And it keeps copies of emails so if Poco loses the body I think I can get it back from K9

AND

by simply moving the [Spam] tag to the start of the subject line and requesting the spam rating to be added I can easily sort my spam in Poco by rating (least likely to most likely) which I think makes it easier to check spam for good mail.
Alycat
Frequent Visitor
 
Posts: 79
Joined: Sat Nov 20, 2004 9:03 pm
Location: Sydney, Australia

Postby ambkla » Fri Dec 03, 2004 11:54 am

Alycat wrote:I am very hard to please...but WOW!


If you want, I can send the txt file I used as "blacklist" containing mostly regexp conditions. This and the "whitelist" kept the spam success always above the 98%; sometimes 100% for several weeks.

Yes, it is very nice that K9 keeps the-mails and renaming anyone of them from .kml to .eml allows importing. The only problem, the e-mail program must be running constantly to perform mail checks.

Klaus
ambkla
Poco Tourist
 
Posts: 20
Joined: Tue Nov 30, 2004 4:13 am
Location: Oregon

Postby Alycat » Fri Dec 03, 2004 1:45 pm

ambkla wrote:If you want, I can send the txt file I used as "blacklist" containing mostly regexp conditions.
yes thanks, poko -at- alycat.com

I have received about 30 emails this morning, ALL of them correctly id'ed by K9 :D
renaming anyone of them from .kml to .eml allows importing.
will have to remember that
The only problem, the e-mail program must be running constantly to perform mail checks
K9? Here, it is only 7.4 Megs, my laptop has a Gig, no worries :D

[ EDIT: "hid" email address from spambots - Pete ]
Alycat
Frequent Visitor
 
Posts: 79
Joined: Sat Nov 20, 2004 9:03 pm
Location: Sydney, Australia

Postby Alycat » Fri Dec 03, 2004 6:23 pm

138 emails received today, 100% accuracy :D :D
Alycat
Frequent Visitor
 
Posts: 79
Joined: Sat Nov 20, 2004 9:03 pm
Location: Sydney, Australia

Postby speerga » Sat Dec 04, 2004 2:32 am

Yeah, if you look through threads here at the Poco forums, you'll find a lot of "yea" and "nay" mixed results for Pocomail's spam filtering.

In my opinion, it's the one real weakness of the program. Will always seem incredible to me that Bayesian filtering, which is so absolutely outstanding in K9 and Spamihilator, can be so BAD in Pocomail.

Currently, I'm relying on Poco's filtering without K9 -- which I used for most of a year -- because I find noticeably slower email downloads running any third-party spam filters with Pocomail. What I'm finding is that most of my spam comes to one or two email addresses, and I can pretty effectively filter it out with a couple of specific filters.

Having said that, there's not a day that goes by when I don't get at least one or two spam messages which slip by the filtering I have set up. And, of course, I dutifully mark them as junk and move them to Junk. Yet it seems NEVER to train Poco's spam filters so that they catch those emails or similar emails later???

For the record -- I currently have 42,312 junk words and 34,433 good words, running at 98% accuracy. The catch is this: If I turn off the two filters I created to catch my two worst spammed email addresses and just use the other Pocomail junk filtering (including the Bayesian filter) that accuracy rate drops to somewhere around 40-60% in a couple of days and NEVER gets any better.

Gary Speer
speerga
Resident Poster
 
Posts: 116
Joined: Wed Jul 28, 2004 1:49 pm
Location: Springfield, Missouri

Postby Alycat » Sat Dec 04, 2004 8:14 am

I have only used Poco for a few weeks, and it could be argued that I may not have given the filters enough time. I had the same experience of marking what appeared to be the same spam email over and over again to no apparent affect.

My experience with all the other email clients I've used that had bayesian filters were mixed but all better than mine with Poco.

With K9, I have now about 350 emails received, and since that intial batch, it has missed one (and I had to have a good look at it to see that it was actually spam) with no false positives. I am using K9 "out of the box". Very impressive :D
Alycat
Frequent Visitor
 
Posts: 79
Joined: Sat Nov 20, 2004 9:03 pm
Location: Sydney, Australia

Re: Spam Blockers

Postby Eric » Sat Dec 04, 2004 1:11 pm

J-Mac wrote:Some use Popmail also.
You mean PopFile, right?
Alycat wrote:I am using K9 "out of the box". Very impressive Very Happy
K9 impressed me too. Howevever you still have to switch twice to see the messages. Once in K9 and afterwards downloading in Poco. :?
Spamihilator works in the background, so the only thing you've to do is check the messages marked Spam. Had some good results in my previous posted test. :)
My previous spam blocker 'Spam Blackout' got extreme good results, but no longer needed, when running at 99,18% with Poco. (and climbing)

Tried/bought enough already in the past to eliminate these pests. :lol:

It's a pity it doesn't work for you all, like mine does. :cry:
Eric
 

Re: Spam Blockers

Postby Alycat » Sat Dec 04, 2004 1:53 pm

K9 impressed me too. Howevever you still have to switch twice to see the messages. Once in K9 and afterwards downloading in Poco.
Only if K9 gets it wrong :lol: The way I see it I haven't burnt any bridges, K9 is going well but if it doesn't work out, I can try something else
Alycat
Frequent Visitor
 
Posts: 79
Joined: Sat Nov 20, 2004 9:03 pm
Location: Sydney, Australia

Postby vamp07 » Sun Dec 05, 2004 12:25 am

Spamihilator is interesting but I really don't get it. It seems to involve too much work. You need to switch to it to see that nothing was thrown away by mistake and you also need to switch to it's training area to learn/clear it out. It's cool that it keeps all spam at it's end but it offers no search to make sure something you did not receive did not mistakenly get misclasssified (try looking though 1000 headers when you know a message you were expecting would contain word x). It also seems to force you to over train. It's my understanding that with Bayesian filtering you should only train on errors. In spamihilator you need to train on everything in the training area to get it to clear out. Make sure that when you click learn you made no mistakes because there is no going back after that. I do like their UI.

My preferred Bayesian filters are:

#1 The built in poco Bayesian filter if it worked.

#2. Popfile if I could get it to not hang after about a day of use.

#3. K9 which works great and is only a close second to popfile in usability (with popfile you can link from poco to messages in popfile using urls).

All spam with k9 or popfile get marked as spam with a special header X-Text-Classification: (no subject modification). If it has a value of spam I use a filter to move it to a folder called spam. Once a day I review that folder and hit the delete key on everything after giving it a quick look over. This moves it to the trash folder. I then clear out the trash folder of anything older then a month on a periodic basis. I should probably write a script to do this. This method of spam filtering lets me use pocos search to occasionally look for mail that I never received and that in my quick spam folder review I may have missed.
vamp07
Frequent Visitor
 
Posts: 66
Joined: Mon Jul 26, 2004 11:31 am

Postby speerga » Sun Dec 05, 2004 3:54 am

I would agree that the best solution would be Pocomail's own Bayesian system. But, as I and others have said repeatedly, the little thing just seems pretty "learning impaired" as far as Bayesian filters go. :lol:

I prefer K9 of the others mentioned. Popfile seems to run somewhat sluggishly in my system. K9 noticeably slows email checking, but not too badly.

The beauty of K9 (perhaps popfile, too, I dunno?) is that it was very quickly trainable, I can go right back to it when I wish, or I can continue with the multi-filter system that seems to be working pretty well for me right now within Pocomail.

Now, if Pocomail's Bayesian filter would just get as good as K9's, or Thunderbird's, or Mozilla Mail's -- I'd be happy!

Gary Speer
speerga
Resident Poster
 
Posts: 116
Joined: Wed Jul 28, 2004 1:49 pm
Location: Springfield, Missouri

Postby ambkla » Sun Dec 05, 2004 4:42 am

[quote="vamp07"][/quote]

Did you ever look at the 25 plug-ins offered for free on Spamihilator's Web site? I have installed: Domainfilter, Countryfilter, Misspelledfilter, Spacefilter and X-Header filter. These filters need no training at all and remove 95% of all spam mails. The rest (5%) is then scanned by the Wordfilter and then by Bayesian.

If you know that an e-mail is coming in, why is the sender or domain not on your Whitelist? Whitelisted mail will never be checked. As soon as it arrives, Spami's pop-up message will let you know provided you installed the POP3notifier.

Take around one hour to set-up and configure Spamihilator and you will see what I mean.

Klaus
ambkla
Poco Tourist
 
Posts: 20
Joined: Tue Nov 30, 2004 4:13 am
Location: Oregon

Postby dribnus » Sun Dec 05, 2004 6:42 am

I used to really question PM's Bayesian system.....however, after much playing around with it, I have been able to get it to stay at about 98-99%.
Here are the steps that I took to get it to this point, as far as I know, these have not been mentioned in the forums. I also have a check mark in the "Run Standard Non-Bayesian Filters"
1. Any spam that is missed, classify it as 'junk' and move to 'junk folder using the 'file as junk' icon
2. Any 'GOOD' message that is sent to junk, right click on it, and, click on 'declassify as junk', then 'classify as good', after that, drag it back to the appropriate folder.
** 3. I think this is the important part....after you have many emails, at least 100, in the 'junk' folder, Highlight all emails, and then press "CTRL + F4", click on 'JUNK' Multiple times, @15.
4. While all emails are still selected, right click on messages, choose "Junk Mail Filtering", then 'Tag Junk Messages', after that choose 'show untagged' in the 'show...' dropdown menu.
5. Everything this is untagged, is still NOT classified as 'junk', and you will need to repeat the 'classifying' steps above, until they are consdidered 'Junk' when you do the 'Apply and Test' step.

I just know that after I took these steps, and repeated them about once a week, I don't have to anymore, and I'm running about 98-99% accuracy with 0% 'false positives'
dribnus
Poco Enthusiast
 
Posts: 221
Joined: Fri Jul 30, 2004 3:33 am

Postby ambkla » Sun Dec 05, 2004 7:38 am

Bayesian filtering is work-intensive and slows down e-mail checking and email downloading. Therefore, it should be used as last possibility to see whether or not an e-mail is good.
For me, it is very important that a program supports regular expressions or similar filters and that defined conditions are easily accessible for re-use after changing an application.

Let's assume you use "MyEmailer" operating with Bayesian and "MyEmailer" using mostly regexp and/ or special filters, you will see that the second program needs maybe 1/4 of download and filter processing time compared to the first one.

Klaus
ambkla
Poco Tourist
 
Posts: 20
Joined: Tue Nov 30, 2004 4:13 am
Location: Oregon

Next

Return to Email Hall

Who is online

Users browsing this forum: No registered users and 0 guests

cron