SpamNet
Cloudmark SpamNet is a large collaborative spam-fighting community. This is an online
database that contains the control sums of text and attachments of spam and phishing
messages, added by the users of SpamNet.
For classifying the messages the plug-in connects to SpamNet servers via internet.
The classifier assigns spam score 100 to a message in case if the database contains
the control sums of message contents, added by the trusted users of SpamNet. The
control sum is a short sequence of symbols, constructed using one-way encoding algorithm
from the random parts of message text and attachments. It is not possible to reconstruct
the original message contents using the control sums.
The speed of checking mail via SpamNet depends on the network throughput and the
current loading of SpamNet servers. Thereby this method is used after all other
methods, in case if the plug-in is unable to classify a message as spam or ham with
high probability. The plug-in uses SpamNet for checking the messages having spam
score from the specified range (between 20 and 85 by default).
This method of blocking spam is very reliable, with almost zero probability of false
positives. It is possible to turn on auto-learning on the messages found in SpamNet.
After training on recognized spam the similar messages will be classified with a high spam score. Therefore
searching the online database will be unnecessary and the classification performance
will grow.
The plug-in uses port 2703 for connecting to SpamNet via TCP protocol. If this port
is blocked by a firewall, the plug-in can connect to SpamNet using HTTP proxy, which
supports HTTP tunneling (it is possible to set the proxy parameters on Options tab of the plug-in settings). Turn off checking the messages via SpamNet in case if the
connection cannot be established using the specified methods.
Important notes:
- Some antiviruses add footers to each incoming email message. This alters the control
sum of message text and impairs the filtering quality. Some versions of AVG antivirus
add to each message a footer, which causes the plug-in to filter all messages as
spam. To resolve this issue, you must configure antivirus so that it does not add
footers. The antivirus will continue to check for threats in this case. The message
headers are not used for classification via SpamNet. Therefore changing message
subjects and adding custom headers doesn't affect the filtering.