POPFile Reviews

A hidden gem

about POPFile · ·

Foreword:

POP and IMAP are not outdated, but they have definitely taken a backseat to apps and web-mail interfaces in recent years. Nowadays virtually every email provider offers some kind of spam protection, which can be a problem if you are like me and want to process all your email traffic centrally (usually on the desktop computer). Multiple addresses and different anti-spam technologies do not only make for very fractured training systems, but some tools (especially those from free mail providers) can hardly be trained at all to your specifications.

That's where a dedicated software is meant to plug in. For many years I have been using the long abandoned Regula Anti-Spam Plugin for The Bat! email client (last updated in 2007) and I was content. The advantage: I could classify mail as spam or not directly from within the email client and the Bayesian filter worked very well; no it worked so well, that I didn't even have to use white- or blacklists to retain a 99.7% success rate after it had been trained properly. Then I switched to Windows 10 and the 64bit version of my mail client and I had to look for alternatives. I tested several different possibilities (SPAMfighter, Spamihilator, Mailwasher, Cloudmark DesktopOne and Phalanx), but ended up choosing a tool with the unassuming name POPFile.

Review:

Unlike some other spam filters, POPFile does not provide a direct integration into the email client. If used with POP3 accounts, it works like a proxy server that is put between the mail client and the server, which means that yes the mail client settings will have to be reconfigured once. If used with IMAP accounts, it will work alongside the mail client and will keep checking the watched folders independently as long as the computer POPFile was installed on has power and Internet access. POPFile does not even need to be installed on the same computer your mail client is running on, instead it's possible to set it up on a server in your local network (that local part is potentially very important, because the first leg of any POP3 connection to POPFile itself is unencrypted) and use it for all devices like your PC, laptop or tablet. In fact, that's what I'm doing with an IMAP account which gets spam-filtered from my home server automatically while I can access it from anywhere on my phone.

Any other spam filter I tested works on the "basic" but nevertheless useful principle of categorizing mails as ham or spam. That's not bad and it is probably everything the majority of users want, but POPFile goes a step further. It offers so-called "buckets", which are user-defined categories that can do exactly the same but have the potential to be so much more. You want separate categories not only for spam and non-spam but also for certain types of content like mailing lists or newsletters? With this tool you can do that!

Once POPFile has been trained well enough - a good guess is several hundred mails but with the more buckets you use the more mails it will take to achieve a good success rate - you can then start to filter your mail either by the subject header modification "[spam] subject" or by the (hidden) "X-Text-Classification: spam" header. These changes to the original mail are configurable per bucket, allowing you to keep the "[spam]" prefix in the subject or not, while using the hidden classification header on all other categorized buckets.

I have found The Bat! to be the perfect companion, because it offers an amazing set of tools for mail filtering purposes out of the box, but with POPFile it can become even better. This even includes a sort-of integration, through which I can reclassify any received mail with only 3 mouse actions. The way to do that: first activate the "X-POPFile-Link" header on all buckets, second go to The Bat! preferences - Viewer/Editor - Message Headers and add said header with the checkbox "Display the field on the scrollable part of the header pane" checked. Then, once you receive a new mail, you scroll down on the header part of the message to see the POPFile-Link url. Double-click that if you want to reclassify the message to another bucket, which will then open the message in the web-interface in your browser. Of course, POPFile can be used with any other mail client as well, like mentioned above it's even possible to be used remotely, with several devices and/or users on the same trained database.

The POPFile web-interface is where you set up or change your buckets (categories), reclassify emails and set up the tool in general. It looks outdated, doesn't offer the most visually appealing or easily accessible statistics and (at least on Windows) hangs on occasion if you're downloading mails at the moment. Any criticism about the lack of a modern interface doesn't matter to me because everything works perfectly well and POPFile exceeds the functionality you could want from a spam filter, but with the way the interface keeps freezing (see below for details) you kind of wish for a new version anyway. In fact, that update has been on the horizon for several years now, but aside from an SSL fix in 2015 the development has stalled since 2011.

More info about the freeze: due to the outdated development status, the use of SSL for connections to mail servers can only be done in a single-threaded mode. If you have several mail accounts and use encrypted connections, they can only be accessed one after the other and not simultaneously. In addition the web interface remains inaccessible while any connection to a mail server is active.

Truthfully, the freeze issue and the fact that you need to install the SSL fix separately are not a big deal at all, but they're definitely worth mentioning. POPFile is not the most user-friendly or accessible, but if you know your way around computers then it is one of the most helpful email tools I know of. In my several hours of testing I'm already starting to see very good bucket classification rates, as is to be expected from any Bayesian filter method. My set-up uses 6 buckets, the extra categories are to separate forum/blog reply notifications, newsletters, mailing lists and business related mail next to regular and spam categories.

Reply