Since we started, Internet Connection has been extremely pro-active at finding solutions to the ever increasing spam problem. Just over six years ago, we made changes to our mail sending policies and software. These changes improved the overall quality of service to Internet Connection customers and benefited the Internet community at-large by implementing an SMTP (mail sending) protocol that had previously only been supported by commercial email clients and server.
A year ago, we launched a powerful spam-blocking and virus-filtering program called Svfilter. Spam traffic was been greatly reduced and customer response has been very favorable.
Still, the most frequent question we've received is how to further reduce the spam count. To answer this, we have given the SVFilter an overhaul and new features. SVFilter now supports a more personalized message classifier that learns with each message that your mailbox receives. Furthermore, errors in classification can now be easily trained on an individual, per-mailbox configuration.
To take advantage of these improvements, you will need to train the SVFilter on any errors that it makes by continuing to use the Retrieve and Whitelist functions of your SVFilter report, but also by letting the SVFilter know when it has let a spam message sneak through. To do that, using your email client, simply forward the message to:
where "user" is your mail user name and "YourDomainName.com" is your domain name, of course.
In addition to statistically learning your emailing habits, SVFilter now adds a unique ID to the bottom of all messages delivered. This signature is used by the SVFilter to identify messages later when it becomes time to reclassify or train them. The message ID will look like this:
This signature only appears when using the SVFilter, and is necessary because the gross majority of email clients in use simply cannot bounce a message without modifying it in some way.
To learn more about these signatures and how the language classifiers actually work, please visit:http://www.nuclearelephant.com/projects/dspam/
If you have any other questions about these signatures, please contact the Internet Connection Support Desk.
Normally, SVFilter can take up to a week or two in order to become trained. This period can be significantly shortened with a two or three day aggressive training exercise.However, making mistakes here can degrade the effectiveness of the SVFilter significantly, and even generate a large number of classification errors that may require contacting our support department for repair. It is highly recommended that you contact our support department if you have any questions after reading this section.
The aggressive training process requires messages of a unique subject matter to be forwarded to each of the following special email addresses, regardless of whether or not the SVFilter made a mistake in the SVFilter's initial classification of the message:
Each study- operation is used to tell the SVFilter about your interest in the message subject matter during aggressive training. It is not helpful to send the same message or messages of a similar subject matter to the study- operations.study-spam is for messages you consider SPAM, while study-not-spam is for messages you feel are legitimate.
Both have to be used (nearly) equally in order for the aggressive training to be effective.
For example, when you forward that email about how you "Qualify for a new mortgage" to user-study-spam@YourDomainName.com you should also take that message you got from your neighbor about their "36 hour summer barbecue next week" and forward it to user-study-not-spam@YourDomainName.com to keep the training system balanced.
When you receive your next message about that "Lowest Viag%ra Prices Online!" forward it to the user-study-spam address, and make sure you send a sales message you got from a new customer to the user-study-not-spam@YourDomainName.com address.
Finally, aggressive training should not be used past the first 50 messages that you receive, or 5 days, whichever comes first. When performing this training, make sure you forward messages to both the study-spam and study-not-spam equally. If you find yourself unable to use both study- operations equally, you should temporarily stop aggressive training and use the more casual training operations.Again, if you have any questions about this process, please contact the Internet Connection Support Desk before you attempt it.