Logo
Go


Step1: Identify support Step2: Issue Step3: Solution and contact

How to configure the Bayesian filter

 

Issue:

The Bayesian filter improves the detection rate of BitDefender's Antispam module by being able to learn how to detect spam messages based on specific examples of e-mails (all relevant words present in the email messages are taken into account). In order to efficiently train the bayesian filter you need at least 100 spam messages and at least 100 valid messages prepared from various users' mailboxes.
The current article presents how to use the Bayesian filter in order to improve the spam messages detection rate.

Solution:

In order to configure the Bayesian filter you need to follow the steps below:
  1. Obtain the email samples for training purposes - please note that our product supports training from .eml format only. Thus please make sure that you save the email messages accordingly. Additionally make sure that you split the messages into two different folders for the mentioned categories (legitimate and spam);
  2. Open BitDefender and click Antispam;
  3. Click Policies - you can either create a new policy for the bayesian filter or use an existing one;
  4. Click Properties;
  5. Go to the Bayesian Filter tab and check Enable Bayesian Filter;
  6. Make sure to specify the paths to the folders containing the legitimate messages (HAM folder) and spam messages (SPAM folder) accordingly;
  7. Provide the training interval in the corresponding edit box. The training interval needs to be high enough to allow the Bayesian filter to finish the training process without impacting on the overall server performance. The recommended time interval is 60 minutes;
  8. Click OK.


We strongly recommend to train the bayesian filter periodically. As soon as the filter has finished the training the email samples will be automatically deleted.



Software Applications:
BitDefender Security for Exchange, BitDefender Security for Mail Servers (Windows)

Operating systems:
Windows 2000, Windows 2003

Solution ID: 402 Created on 23 Oct 2007 12:27
Language: English Modified on 23 Oct 2007 13:58


Useful Links


MyBitDefender Login