Thunderbird: Smartest Way to Stop Junk Mail

Thunderbird: Smartest Way to Stop Junk Mail.



Standard Disclaimer: This page has been designed as a simple guide for installing Thunderbird 1.0 on a Windows machine, especially for viewing mails from the mail server at Indian Statistical Institute, Kolkata. Many of the instructions may be equally applicable elsewhere. However, the definitive guides for this purpose and more information are available here:
http://www.mozilla.org/products/thunderbird/
http://www.mozilla.org/support/thunderbird/
http://texturizer.net/thunderbird/faq.html
Having said that, I am pretty confident that by following the instructions provided here, you shall be done with the installation without any difficulty.

Thunderbird 1.0 is a marvelous e-mail client that may be used as a replacement for Outlook Express, Netscape or any other email client. There is a Windows installer version available on our WWW machine, or you may download the setup file from mozilla.org. The biggest reason in favour of switching to Thunderbird is its advanced junk-mail filtering technique, namely Bayesian filtering: "Thunderbird has several ways to detect junk mail, or unsolicited mail. ... It can be trained to analyze the contents of your incoming messages and identify those that are most likely to be junk. This can be an extremely effective tool for detecting junk mail."

The following is an informal guide on how to set up Thunderbird and activate the junk mail filtering. Well, since your original emails are not being destroyed, if not satisfied, you can always switch back to your original email client. Please do not uninstall the existing client unless you are completely satisfied and feel comfortable with Thunderbird.


First a few things that Thunderbird (or at least this guide) can't/doesn't do:

ThunderBird cannot import/handle folders with special characters in their names. For example, if you have a '/' in your folder name, go to Outlook Express and rename it there.

Thunderbird cannot access your hotmail accounts (I am not sure about this one, but read about this somewhere).

Managing multiple identities is something that I am not certain about. However, if you have that in your Outlook Express and you want to import it that way, you may have a look at this information about managing multiple identities.


A few other things before you start installing

If you want to run both your present email client and Thunderbird parallelly for a few days for the sake of evaluation and comparison (which is indeed recommended), you need to change the settings in your current client to "Leave messages on Server".

If you have chosen to leave messages on the server (for manual deletion, perhaps), Thunderbird would download them when it is first launched (because it would not know that these are duplicates of the mails that your other email client has already downloaded). Hence you would be having duplicates of those mails (one copy having been in existence after being imported from the existing email client). You may either delete them from the server itself or delete the duplicates after Thunderbird downloads them.

Training Thunderbird for junk mail detection needs you to show it a few mails of both junk and non-junk varieties. So, please do leave some junk mails in your existing mail client.


Installing Thunderbird 1.0

You can now proceed to install Thunderbird 1.0. A few snapshots are provided below, along with the instructions, for your convenience. You might want to read all the instructions first so that you may choose the options best suited to you.

1. Double click the Thunderbird setup file.


2. Click Next and click on "I Accept" (Why is it there at all ?).


3. If you feel confident, you may choose a custom installation (instead of the standard one), mainly to change the default directory in which Thunderbird installs itself. For example, one might want to have the installation in D: (instead of the default C:). The default installation should be just fine. Keep clicking Next, and this should take you through the installation, and you will reach the "Install Complete" dialog in a minute.


Click finish. If you have not unchecked the "Launch" button, Thunderbird will be opened now. Otherwise, open it manually. Do not enter your password (even if prompted to do so) until you are finished with the setup.

If this is the first time that Thunderbird has been opened here, it opens an "Import Settings" Window. Choose the appropriate email client (the one from which you are shifting... most probably Outlook Express or Netscape Communicator). I had noticed that on some Windows XP machines, the items being imported are not shown. However, the importing itself was fine.


Thunderbird should now have imported everything from Outlook Express. Click on each of the folders and sub-folders to ensure that all your mails have been imported. (It is not too difficult compared to amount of spam that you will be spared from later).

You would notice that all your mails are marked Unread. You can mark them Read by pressing "Ctrl+Shift+C" for each folder.


Though it looks like we are done, it isn't so. There still is some configuration remaining to get your ISICal mails. Now proceed to Account Settings.


Note down the folder Thunderbird has chosen to store the emails. (The path mentioned near the "Browse" button. It is "C:\Documents and Settings\new\Application Data\Thunderbird\Profiles\b7lckipx.default\Mail\" in my case). If you really need to, you may move this Mail folder to some other place manually and then click the Browse button and set it to "Local Folders" there. Again, it should be just fine if you go with the defaults.


Other than the Local Folders, there's an account for "www.isical.ac.in" (see figure below). This may vary with the settings of your earlier client (for example, it might be 202.54.54.145).


In the Server Settings, check "Leave messages on server". This is so that messages will remain on the server and you can access them parallelly through your earlier email client for the sake of testing and comparison.

Now choose "Outgoing Server (SMTP)".


Change the default port from 25 to 500. Also, uncheck "Use name and password". Click OK and your setup is now complete. Send a test mail to yourself and see if you can both send and get it.


Now for the much awaited Junk Filtering.

Hope you have not deleted all your junk mail. If you did, just wait a little and you will have one soon. The column between "Sender" and "Date" denotes "Junk Status". For each junk mail (and only for the junk mails), click (see below) to mark the message as junk. Thunderbird will respond with a message saying "Thunderbird thinks this is a junk message".


This is the training part. Now, wait for more mails to come. Observe if Thunderbird has automatically begun marking any new mails as junk. If it has marked any message incorrectly as junk, rectify it by removing the junk status. Pretty soon (how soon depends on the number of mails as well as the variety of spam that you receive), you should observe that Thunderbird makes fewer and fewer (or almost no) mistakes.

It's now time to set your Junk Mail Controls. Choose it from the Tools menu. You should now see a dialog box like this one:


Set appropriate controls like deleting or moving junk mails to designated folders.

If you feel that Thunderbird is doing really bad with the filtering (which might be a consequence of poor training), you can reset its parameters. Choose "Adaptive Filter" and choose "Reset Training Data". You now need to retrain Thunderbird. (It is not necessary that you train it straightaway. You may do so as and when you receive junk mails).


A few other niceties

Other than the junk mail filtering, you can also try out some of the following:

Explore Tools-->Message Filters. Just imagine that you are writing a program, with lots of if-then statements, to decide which folder an incoming mail should go to. This is just the equivalent of that.

Choose Tools-->Options and adjust the settings to your taste. In particular, some of the options under "Advanced" my be of interest.

Under Tools-->Account Settings-->Copies and Folders, you may bcc yourself (the same email id), and then use message filters to put it into an appropriate folder. That way, instead of having all your sent mails in one folder, you can keep them where they actually should be.


That's more or less it. Of course, if you find any errors, please inform me by email (bln_r), and I shall (employ Bayesian Learning to) rectify them.

Thanks to Sarif Kumar Naik (SRF, MIU) for help in creating this guide, and to several volunteers for helping me in the testing phase.


B. L. Narayan
Machine Intelligence Unit,
Indian Statistical Institute.