Keep Spam At Bay With SpamBayes

Published Date
01 - Mar - 2005
| Last Updated
01 - Mar - 2005
 
Keep Spam At Bay With SpamBayes
We have mentioned Bayesian spam filters before in Digit. Here, we give you a 30-minute brief on the most popular Bayesian spam-guard called, very innovatively, SpamBayes. The way it works is, the program, using Bayes' theorem, assumes that each message has evidence for being spam as well as ham, or 'not spam'. Every time you see a message, you ask yourself about it: "Is it spam or ham?" Then, what you do with the message answers the program's question one way or the other, and the system learns. The program thus learns according to how you deal with messages. So, if over time, your preferences change, so will your spam settings-and that too, automatically!

Step 1. Get it
To set up SpamBayes, you will need to download the installation file from http://snipurl. com/cupc.

This is a complete installation file and will work on Outlook 2000/2002(XP)/2003. It will not work with Outlook Express.

Hold it! Once you have this downloaded, don't just install it. There are some preparations you need to do in order to make SpamBayes function.

Step 2. Pre-Installation Preparation
SpamBayes learns by example, so you need to show it a set of good mails (ham) and a set of bad mails (spam). There are probably good mails residing in your Inbox right now. And there is probably some spam in your Junk Mail folder. Just keep these two sets of examples ready. If for some reason you can't set up these two folders, then you will just have to let SpamBayes learn as you classify and delete mails as and when you check them.

Step 3. Installation
When you install SpamBayes it will ask you to choose from two options: Microsoft Outlook Addin or Server/Proxy application, which is for all other (non-Outlook) POP3-based mail clients.

You can choose either or both at this screen

SpamBayes will appear as a toolbar when you start Outlook.

Step 4. The Training
After the installation, the first time you start Outlook, you will get a pop-up asking you whether or not you have the folders configured as mentioned in step 2, or if you would rather let SpamBayes learn. If you do have the folders configured, simply point SpamBayes to them and it will analyse all your e-mails, and create its own set of criteria that will govern where future messages go.

The SpamBayes toolbar

If you do not have the folders configured, then just check your mail as you normally do, and when you want to classify something as spam, just select it and hit the button marked 'Delete as spam'. The message will be moved to your Junk folder. If you accidentally delete a message that is ham, just select it and click 'Recover from Spam'.

To edit SpamBayes settings, click SpamBayes, and then SpamBayes manager. Here you can see how many messages it has on hand to learn from. Obviously, the more the messages, the better the filtering.

Since the program just awards points (probabilities) to messages to classify them as spam or ham, you can edit the points and the folders associated with those points via the Filtering tab.

The training tab allows you to specify the folders from where the program will learn.

Configuration For Non-Outlook Users
If you are using a different e-mail client such as Outlook Express or Eudora, then
  • Install as option 2 (that is Server/Proxy application), as specified above.
  • Right-click on the icon in the system tray and click configure. You will come to a Web page-based settings menu.

The web-based manager for SpamBayes

Here you need to enter your e-mail server settings and the requisite ports. Generally, use port 110 for the POP3 server and 25 for the SMTP server.
  • Once this is set up, you need to go to the connection settings for your e-mail client and set the POP address to 127.127.0.0. If for some reason this doesn't work, then just type in 'localhost'.

The setting page for Outlook Express

Now every time your mail comes in, SpamBayes will analyse it and classify it a ham, spam or 'unsure'. You can configure it to add the classification word in the subject line of your message. Or you can create filters to direct the mail into the folders accordingly. You can correct improper message classifications via the Review Messages page, the link to which is found on the home page.

What remains to be seen is how good the system will be at filtering out spam of the future, and whether spammers can come up with something innovative enough to beat SpamBayes. But for now, you are set!



Team DigitTeam Digit

All of us are better than one of us.