Author Topic: How to accurately block junk mail? (The Bat and other clients also) (Read 33,871 times)

superboyac · « **on:** December 11, 2006, 03:27 PM »

I'm using the Bat, and I have several addresses. On one of my addresses, and this is rare for me, I'm getting junk mail occasionally. I'm using the Bayesit plugin that is supposed to handle this stuff pretty well. But it's one of those junk mails that generates completely different email addresses for the sender (the prefix and suffix of the address is always totally random). And the message contents are just random words put together, gibberish. I've been marking the emails as junk and not junk in order to "train" the plugin and doing it for almost a year. However, no matter how much I train it, it never seems to be able to catch anything. Is there any way to catch these completely random emails?

KenR · « **Reply #1 on:** December 11, 2006, 07:32 PM »

Hey Superboyac,

It sounds like some of the mail that you are getting is pretty similar to what I get. I have been using AntiSpamSniper for the Bat (http://www.antispamsniper.com)with good success. At first, I wasn't getting any benefit. The developers told me a couple of things to change though and now it catches most of the junk by far. Good Luck!

Ken

superboyac · « **Reply #2 on:** December 11, 2006, 09:50 PM »

Thanks KenR, I will try this out tonight.

But does it do anything that the Bayesit plugin can't do, or is it just easier to set up? Supposedly, the Bayesit plugin can also be trained to work, I just can't figure out how to do it.

superboyac · « **Reply #3 on:** September 12, 2007, 04:40 PM »

You know, After all this time, I'm still not catching anything with the Bayesit plugin for the Bat. What the hell am I doing wrong? Every time I get spam, I mark it as junk...yet not once has the program automatically caught a message. I keep having to mark it. I don't get this antispam stuff, I don't think it's ever worked for me. Even at work, with our Groupwise mail, we have a junk filter. Every time I get a spam, I'll mark it as junk, and has it ever caught anything on its own? No. I'm a savvy computer guy, why can't I figure this out?

Carol Haynes · « **Reply #4 on:** September 12, 2007, 05:01 PM »

Try POPFile - let's you filter any way you like but learns your SPAM rules v. quickly as one method of sorting. It uses Bayesian filtering methods and is excellent. See http://popfile.sourceforge.net/

You can use it with any POP email client.

f0dder · « **Reply #5 on:** September 13, 2007, 10:01 AM »

BayesIT worked fine enough for me, but it was slightly unstable (sometimes I couldn't close TheBat, and sometimes mail receive would timeout in BayesIT code). I've been using (the free version of) AntiSpamSniper for quite a while, and it works a lot better... once you've configured the plugin, you need to "train" it - mark junk mails, select "specials->mark as junk". Pay attention to your "junk mail" folder, mark things that aren't spam as "specials->mark as NOT junk".

Here's what my detection rate currently looks like:

Carol Haynes · « **Reply #6 on:** September 13, 2007, 10:10 AM »

Just to compare ... here are my POPfile statistics:

but remember POPfile knows nothing about SPAM at all (and doesn't come with any pattern files etc) it just learns what YOU think is SPAM as it works.

To be fair I reset the counter on the date stated in the stats - the software had been installed for about two weeks at that point - so it isn't doing too badly!

Note on the right hand side the term 'buckets'. These are defined by the user and you can set up as many as you like. The concept is very simple - imagine post coming through the letter box - mail is sorted automatically into the buckets. If one ends up in the wrong bucket you simply say which bucket it should be in and the software learns a bit more about what you mean by your preferred bucket contents.

superboyac · « **Reply #7 on:** October 23, 2007, 03:16 PM »

OK, I tried AntiSpamSniper for the Bat, and it worked great for the last few weeks. So I bought the pro version. It's a great program. Other programs probably work fine also, but I liked the fact that this program is integrated with the Bat (plugin, toolbar, etc.). It works very well, and I LOVE that I have a detailed log of what it does.

Ralf Maximus · « **Reply #8 on:** October 23, 2007, 04:31 PM »

One thing I love about PopFile is its ability to define override rules ("magnets") that bypass the filter for mail you KNOW isn't spam. So even if my goofy sister sends me something that looks like spam it still makes it through.

Tekzel · « **Reply #9 on:** October 23, 2007, 04:39 PM »

At home I used Thunderbird, and I find its built in spam filter works VERY well for me.

At work I have to use Outlook 2003 (ugh) but I use Spambayes to keep that under control there, I have been very imporessed with it's performance over the couple of years I have used it. It also has a "proxy" mode so you can use it with other email clients.

masu · « **Reply #10 on:** October 23, 2007, 04:41 PM »

I still use K9 in connection with the bat! and it works like a charme

hamradio · « **Reply #11 on:** October 23, 2007, 06:04 PM »

Most of mine is done on the server side of things.

Darwin · « **Reply #12 on:** October 23, 2007, 06:37 PM »

Big SpamBayes (FOSS) fan here... use it with Outlook 2003 and only ever see junk mail when I open the folder to check for false-positives. Haven't seen one in months, though I get 20 or more spam e-mails a day over the four pop3 e-mail accounts that I use Outlook to check). SpamBayes also works with other pop3 clients, though I've not tried it with anything else.

tamasd · « **Reply #13 on:** January 12, 2008, 01:00 AM »

I'm a big SpamBayes fan as well. I never had a false positive. I even kept checking for quite a few weeks after the initial one or two week training phase, but no false positive whatsoever.
All in all I save insane time on not having to check junk mail folders anymore.
Yes, I too still get some spam emails, but it takes a second to delete. If I could be bothered with some more training and tweaking, it would for sure go down significantly.

doublewitt · « **Reply #14 on:** January 12, 2008, 02:03 AM »

I am trialing this lightweight (USB install and self-contained) eMAIL client called Espresso. I discovered it recently. It uses a "permissions" based filtering technology. So far - so good. Spam is controlled very good so far. It lets you create assorted "mailboxes" to organize your eMAIL database. I like this so far. Has anyone else tried this software? I wonder; is the "permissions" based filtering more efficient? I only have a 14 day trial, and so, I sort of need more time...

Anybody know more about the permissions vs bayesian?... which generally produces better results...

Josh · « **Reply #15 on:** January 12, 2008, 02:29 AM »

I use Spamato. This combines 6 different filtering systems and allows you to specify how many of them must identify a message as spam before it is marked as such. It's been quite accurate and configurable for me. A great addon!

tamasd · « **Reply #16 on:** January 12, 2008, 10:19 AM »

Quote from: doublewitt on January 12, 2008, 02:03 AM

Has anyone else tried this software? I wonder; is the "permissions" based filtering more efficient? I only have a 14 day trial, and so, I sort of need more time...

Anybody know more about the permissions vs bayesian?... which generally produces better results...

Ask them for more trial time, most shareware authors can help with that.
I wouldn't consider permission based filtering myself, as I would view it as bothering the people that I email with, and other filtering methods work well for me. Yet I can see this method has it's place on the market.
There are people (very small minority) who ignore authorization requests.
You can implement simple permission based system in more powerful email clients too, even if it's not built in, through filtering. For example in Pocomail it would be easy, and I bet some other email clients could do it too.
(Espresso looks neat indeed, judging from the screenshots.)

Carol Haynes · « **Reply #17 on:** January 12, 2008, 11:11 AM »

Quote from: tamasd on January 12, 2008, 10:19 AM

There are people (very small minority) who ignore authorization requests.

Not such a small minority - individuals can cope with it but if you want to communicate with anyone in business or subscribe to any kind of newsletter (even paid newsletters) etc. forget it they simply won't do it because if it became widely used it would be totally impractical. Imagine a company with 100000 customers has to validate every email sent out?

yksyks · « **Reply #18 on:** January 12, 2008, 03:08 PM »

According to my experience with SpamPal, discussed here, there's no reliable way to filter spam with a single method/filter. In my situation the Bayesian filter gets about 80% of spam, the rest is divided between RegEx and collaborative (P2P) filters. In fact the latter using DCC is becoming more and more useful, especially with the spam of type described here.

Carol Haynes · « **Reply #19 on:** January 12, 2008, 03:33 PM »

I have been wondering about chaining POPfile which has really good Bayesian filtering and Mailwasher which has RegEx rule setting but I haven't managed to get POPfile working with other proxy clients (it works fine at the moment as the only filter on my system but I seem to get into hot water when I try to set up all the ports configuration to get one filter to pass stuff to the next filter).

I also have the Yahoo filter check mail before it even gets off the POP server so I have to check my account every now and then only to check the Junk folder doesn't have anything I want. That is really depressing as I am currently getting over 300 spam emails a day. I could just allow it all to go through POPfile but I don't want to wait 10 minutes everytime I check my email while it downloads tons of rubbish.