User`s guide

Retrain Learning Engine (Bayesian)
Before training the Learning Engine (Bayesian), prepare a folder containing only
SPAM messages and another one containing only legitimate messages. The Learning
Engine will analyze them and learn the characteristics that define the spam or
legitimate messages that you usually receive. In order for the training to be efficient,
there must be over 50 messages in each category.
To reset the Bayesian database and retrain the Learning Engine, follow these steps:
1. Open your mail client.
2. On the Acronis Internet Security Suite 2010 antispam toolbar, click the Wizard
button to start the antispam configuration wizard. Detailed information on this
wizard is provided in section Antispam Configuration Wizard (p. 264).
3. Click Next.
4. Select Skip this step and click Next.
5. Select Clear antispam filter database and click Next.
6. Select the folder containing legitimate messages and click Next.
7. Select the folder containing SPAM messages and click Next.
8. Click Finish to start the training process.
9. When training is completed, click Close.
Ask for Help
If this information was not helpful, you can contact Acronis for support as described
in section Support (p. 301).
33.4.2. Many Spam Messages Are Not Detected
If you are receiving many spam messages that are not marked as [spam], you
must configure the Acronis Internet Security Suite 2010 antispam filter so as to
improve its efficiency.
If you are using one of the mail clients Acronis Internet Security Suite 2010 integrates
into, try the following solutions one at a time:
1. Indicate undetected spam messages. This is used to train the Learning Engine
(Bayesian) of the antispam filter and it usually improves antispam detection. The
Learning Engine analyzes the indicated messages and learns their patterns. The
next e-mail messages that fit the same patterns will be marked as [spam].
2. Add spammers to the Spammers list. The e-mail messages received from
addresses in the Spammers list are automatically marked as [spam].
Troubleshooting
296