 SpamFilter
eSoft's SpamFilter blocks over 99% of inbound and outbound spam with low false positives, individual user quarantine controls, and reputation filtering.
SpamFilter will automatically reject, quarantine, redirect or tag and deliver filtered e-mails. To prevent unintended filtering, SpamFilter allows you to create your own "white lists" of trusted senders, such as customers and business partners. eSoft's anti-spam solution incorporates the latest spam fighting technologies, including Bayesian Filtering, Historical Averaging, enhanced Heuristic Analysis, and SPF technology. With the Bayesian Teaching Tool, an Outlook plug-in created by eSoft, you can help SpamFilter learn so that it gets better at catching spam on its own. Additionally, SpamFilter supports Image and PDF spam blocking.
A sophisticated, self-learning Bayesian filter automatically learns how to differentiate good email from spam, phishing attempts and spoofed messages.
“PDF spam, which we were getting inundated with, disappeared!”
— Ken Gander, CCC Computers
SpamFilter's Interface
The graphical user interface allows for granular control of email identified as spam. Each email is analyzed and given a spam score by the various technologies embedded within the SpamFilter engine (Bayesian, heuristics, blacklist, whitelist, reputation, etc.). Spam may be rejected, quarantined, or delivered normally, based on its spam score and the threshold settings within SpamFilter.
Customizable Spam Scoring and Settings
SpamFilter gives you the total control over spam scoring – you can use SpamFilter’s default scoring or you can set custom scoring levels for High, Medium and Low spam.
Graphical Reporting
ThreatMonitor™ provides graphical on email activity, showing spam, virus, clean and other emails by company, individual, with flexible reporting period options.
|
Quarantine Management
Every email that comes through eSoft's SpamFilter gets a score indicating the likelihood that it is spam. Administrators can configure what to do with email depending on its score. Typically emails with medium and high spam scores are rejected while emails with borderline spam scores are put into a quarantine. The quarantine gives the administrator and the addressed user a chance to review borderline spam emails and train the Bayesian Filter on that email. With eSoft's Central Quarantine Management, administrators can centrally manage, release, reject, blacklist or whitelist any email. Individual users also have the power to manage their own quarantines and are optionally notified of quarantined emails in a summary email with embedded links for releasing or deleting quarantined emails.
|
SpamFilter Modules
eSoft fights spam with every tool at its disposal. The modular system uses a variety of technologies to identify unwanted emails including the following:
“The mail security features are extensive. eSoft employs its own proprietary anti-spam measures, which combine a wide range of weapons including Bayesian filters and heuristic analysis.”
— Dave Mitchell, SC Magazine UK
Real-Time Blackhole List (RBL)
eSoft’s SpamFilter includes support for a number of Real-Time Blackhole Lists or RBLs, including eSoft’s Distributed Intelligence Architecture™ (DIA) blacklist. RBLs allow SpamFilter to block email before it enters your network, reducing the load on your email server and other network resources.
URL Black Lists (URBL)
eSoft's Distributed Intelligence Architecture™ (DIA) tracks URLs that are sent inside spam messages. Spammers can hide the content of their message by putting it in the middle of a chapter from a random book, but in the end, they want you to go to their website and eSoft knows spammers' URLs. eSoft blocks emails that have URLs linking to the websites of known spammers.
Bayesian Learning
What's spam to one person is a desired email to another person. For this reason, and because the hard drives in eSoft's appliances give eSoft the ability to remember per-user preferences, eSoft offers lets users train the system on what is wanted and unwanted email. To do this, eSoft keeps a statistical database of keywords and phrases and the number of times those keywords or phrases indicated a message was spam or not spam. This trainable database uses Bayes' theorem of conditional probability to calculate the likelihood of a particular email being spam when a new email enters the system and this information is added back to the other scores returned by other modules in eSoft's SpamFilter.
Out of the box eSoft's SpamFilter catches over 99% of spam. With training, this percentage climbs up to nearly 99.999%.
Heuristic Analysis
Each email message is analyzed against multiple rules and gives it a numerical rating based on which rules were matched. These rules are wide-ranging from the number of words containing numbers and special characters in them (eg \/1agr4 instead of viagra) to indicators that an email was forged to look like it was sent by Outlook when it wasn't. eSoft uses hundreds of these rules with a set of default plus and minus scores that are adjustable by the administrator.
|
Reputation Filtering
Through the magic of eSoft's Distributed Information Architecture (DIA), every subscribed eSoft customer reports the source IP address and spam score for every piece of e-mail. In return they receive the global score for that IP address, which is weighted to take into consideration the amount of spam coming from that IP address. This global score for that IP address is then averaged in to the pre-calculated spam score to come up with a reputation-adjusted score. Since eSoft customers use the Bayesian features to train their spam filter, all eSoft customers get the benefits of that global training in determing whether a particular email should be rejected or quarantined.
eSoft's implementation of reputation filtering is unique in a number of ways including how global scores are calculated and weighted, and the speed and effectiveness of the system at identifying and blocking sources of spam.
Sender ID and Sender Policy Framework (SPF)
SPF is a mechanism for checking that the server sending a particular emails is authorized to send an email from its claimed domain. This check drastically cuts down on the number of forged emails so that, for example, a spammer can't pretend to be sending email from the address support@microsoft.com.
Sender Address Verification
SpamFilter will optionally check to make sure that the purported sender in the from field of an email is a valid email recipient so that spammers can't make up user or domain names when sending email without getting caught. This check is necessary since not all domains support SPF and spammers often attempt to spoof emails from these unprotected domains.
|
| 
