Desktop Spam-filtering Tools

We've all seen spam in one form or the other. If a recent
study is to be believed, spam costs companies over $10 bn a month worldwide in
system downtime and employee time lost that could have been utilized
productively.

While we manually handle spam using the 'delete and move
to next mail' approach,  it becomes
difficult to do so in many cases and one or more tools are often necessary to
keep this menace in check. The job of handling spam can be done either at the
server or the client end. While the former approach monitors and runs scripts on
the server for all/several mailboxes/users that can be done by the system
administrator, the latter is something that can be done by individual users for
their own mailbox.

Let's look at some desktop spam-filtering tools for how
they stack up against each other. But before that, let's understand some of
the key concepts associated with this and the various techniques that these
tools use to classify a mail as spam (or otherwise).

How tools classify
spam
 

One way to classify mail is to have a central server to
store information about known spam sending e-mail/IP addresses. The headers of
the mail are checked against the data obtained from this server and if it
matches, the mail is marked as spam. The server can be contacted every time the
mail is checked, else you can configure software to download information about
known offenders from the central server, say, once a week and keep a local copy
for reference-similar to the virus definitions model. These lists are called
'blacklists', wherein the software knows that the moment a match occurs with
one of these addresses, the mail can be classified as spam.

There are other software that need to be trained to
classify them as spam. To begin with, they have no blacklist to match the mail
against. You have to mark junk mail in your inbox as spam so that the software
learns and then extracts information from  such
mail-like the sender ID and IP address to add to its blacklist. You can also
tell the software explicitly what information to use from that e-mail, like
blacklist this e-mail address, this IP address and this domain.

Perhaps the
best tools are ones that combine these two, and spread what they have learnt
across the community. That is, when a person marks such a mail as junk that the
software used to identify as spam, the tool not only learns locally but also
reports back to a central server so that other users can benefit from their work
when they update their blacklists. Like blacklists, there are 'whitelists'
as well. Whitelists feature trusted senders, mail from whom is not spam. This is
people specific as your set of trusted senders is specific to you. So whitelists
need not be stored on a server or shared with others. Other methods used to
identify spam include inspecting the subject line and the body of the e-mail for
specific words (say, pharma and pills) or phrases and flagging them as spam
accordingly. Again this list can be central, local or both and, as usual, the
local list takes precedence over the central one where both exist. Some tools
also offer the facility to block all mail originating from certain countries or
top-level domains.

Tools also differ in where they fit in during the e-mail
checking. While some require themselves to be given access to the mailbox before
you allow your e-mail client to download mail from the account, other
transparent ones run as proxy servers on the local machine. Your e-mail client
is then given the address of this proxy server which downloads e-mail from your
server, filters them, delivering the good ones to your inbox while the spam goes
to a quarantine folder or to the bin, depending upon the options offered by the
software and your preferences.

Most tools operate on POP accounts, some support IMAP while
others are capable of weeding out of spam over Web e-mail accounts like Hotmail
as well.

Here we put some of the popular spam filtering desktop
tools to test with the emphasis, obviously, being on their ability to weed out
spam from legitimate mail. Solutions that do the job out of the box have a head
start over those that start from scratch and require learning. Tools that offer
both built-in/server-based as well as local/user habits learning based filtering
score higher than those that offer only one. Flexibility is another
advantage-for example, the ability to have user rules over rides built-in
ones. Additional features like language-based filtering mean extra points. But
ultimately everything boils down to how effective the tool is in filtering out
spam with minimum of user effort after the initial training. The process
automation is the key-be it via downloading 'definitions' or filters from
a central server or by using advanced filtering algorithms-because if your
tool requires tinkering every second day then it isn't serving much of a use.
The winner here reflects that philosophy.

yMail
 

www.spacejock.com/yMail.html

yMail is a stand-alone, portable e-mail client with integrated spam-tackling
capabilities good enough to make it one of the contenders in our shootout. The
portable bit means that you can carry it around on a USB drive as it can be
configured to store mail, contact information and the likes in the same folder
as the program files.

The e-mail capabilities are pretty basic and we think
it's more suited to an e-mail popper role to filter out spam and other
unwanted e-mail, meaning it's to be used in combination with your current
e-mail client (like Outlook or Eudora) and not necessarily as a replacement.

yMail uses Bayesian techniques to classify spam and does a
very good job at that. It starts with a clean slate and learns what is spam and
isn't as you go along. This means that you initially have to spend some time
and effort with it, but every mail you mark as spam adds to yMail's knowledge
and increases its ability to identify junk mail accurately the next time. In our
tests when once we marked the first batch of junk mail as spam, yMail was good
enough to weed out all further junk mail on that account-and that is an
achievement good enough to make it the winner in our shootout, as far as its
spam-filtering abilities are concerned.

SpamAware

www.jam-software.com 

This tool is
very similar to SPAMfighter in terms of look, feel and functionality- infact
we had to closely re-examine the two to really appreciate the differences. For
example, its default behavior is to mark junk mail by adding a prefix 'SPAM'
to the subject, while SPAMfighter moves it to a folder it creates. Of course,
you can then create a filter in your e-mail client to move all mail with a
prefix SPAM to trash else you desire.

Like SPAMfighter, it sits inside Outlook/Outlook Express
and automatically scans all incoming mail. Where it fails to deliver is in terms
of performance. The scanning process is a crawl compared to SPAMfighter and
doesn't offer any improvement in terms of accuracy to compensate for
that.It's a free tool and is the only one in the pack that doesn't support
custom keyword filters. It has all the regular features such as support for
whitelistas and blacklists, language-based filtering, etc. It can compare the
headers of a suspicious mail against data obtained from a central server.

PAMfighter

www.spamfighter.com
 

SPAMfighter sits inside Microsoft Outlook and Outlook
Express, and monitors incoming mail on all accounts. It compares the mail
against locally configured blacklists and whitelists as well as rules that are
periodically downloaded from a central server. Mails that are identified as spam
are moved to a separate folder within Outlook/OE, which you can browse later, if
necessary, to ensure that no useful mail was junked and/or clean the folder.

While it doesn't support keyword-based classification of
e-mail, it does innovate in the form of language-based e-mail filtering. For
example, you can configure it to junk mail in languages other than English. It
also provides for you to report (via single click) to the central server, spam
that might have been missed by the automatic detection process.

All in all, it offers pretty good filtering capabilities at
an almost real-time speed via an interface that is easy to use, fitting in
nicely within Outlook/Outlook Express. It requires no changes in e-mail settings
or how you go about checking your mail. Thus, most people wouldn't even notice
it (apart from the good work it does), which can't be said of some of the
other tools that we've seen.

Sure it isn't of any use to those who prefer other e-mail
clients but that shouldn't take much away from the good job it does.

Mailwasher  Free/ Pro
 

www.mailwasher.net

Mailwasher is like an e-mail popper with advanced spam
filtering capabilities. You set up an e-mail account just like in an e-mail
client and it downloads the mail headers for you to preview, while marking the
spam it can identify. The free version identifies spam using built-in filters
and, of course, allows you to manually mark spam. You can then proceed to delete
these messages from server and/or send a 'bounce back' to the sender, which
looks just like the kind of error mail you get back from an invalid e-mail id.

The Pro version offers much better protection from spam by
allowing the client to maintain black and white lists as and when you mark
specific messages as spam and/or those from trusted servers. It offers access to
a global spam database and DNS spam blacklists both of which are maintained
online on central server(s) and greatly increase the effectiveness of the spam
identification process.

Once spam has been marked/bounced/learnt from, you can launch your e-mail client and
download e-mail. You can use almost every mail client. Mailwasher has a button
to launch your default mail client-a small but handy convenience.

The Pro version also supports keyword-based filters,
multiple accounts (as opposed to only one in the free version) as well as access
to IMAP, Hotmail, MSN and AOL accounts in addition to the POP accounts supported
by the free variant. It also allows you to preview a message before you decide
what to do with it in case you are not able to identify the nature of the
message just by looking at the headers.

Letterman Spam
Control Pro
 

www.whatlink.com
 

This is perhaps the
most feature rich tool amongst the ones we gave a run during the shootout. It
can operate in multiple modes-as a standalone 'popper' to filter out your
mail before download which you can use, say, at the start of the day when
you're checking mail for the first time. And for the rest of the day you can
use it as a proxy server, an intermediary between your e-mail client and server
that not only filters out spam but also blocks hostile links in other e-mail. It
does a good job of filtering mail straight out of the box using built-in rules
and allows creation of custom rules.

Letterman Spam
Control also supports filtering of Web-based e-mail accounts such as Yahoo! and
MSN/Hotmail. Blocking of mail based on their language and analysis of mail
written in languages other than English (like Chinese) is also supported. It
also provides detailed logs/reports for each message indicating the criteria
that lead to the message being marked as spam.

All this sounds
impressive but where it loses out is usability. It offers a plethora of options
but doesn't feature the most intuitive of interfaces, which means that it can
be quite intimidating to most users. We had to spend considerable time to find
where each feature option was before we could gain enough familiarity with the
tool.

Kunal Dua