Plugin Modules provide a way in which Spam Annihilator can be constantly updated to catch new tricks employed by spammers to get around spam detection programs. Part of the detection process of Spam Annihilator has been implemented in an internal set of rules and tests using a database of known spammer keywords and phrases. Other spam detection techniques have been implemented in a set of plugin modules. Each time you start Spam Annihilator it can check the Spam Annihilator website and download any new or updated modules. In this way we can keep Spam Annihilator bang up to date with spammers techniques to disguise spam. You can disable this behaviour if you do not want to install new modules.

The Spam Annihilator software comes with a basic set of plugins. The current release status of each plugin is as follows;

• Bayes - a bayesian classifier - beta
• BlockExe - block exectuable attachments - beta
• CharSet - block emails with certain character sets - alpha
• Dictionary - check words exist in a dictionary - experimental
• Garbage - check for garbage words - experimental
• SpamHaus - check IP addresses in SpamHaus Black List - beta

The modules developed for Spam Annihilator have been used to create the Pugwash Open Source project. Using the information from Pugwash you can develop your own plugin modules for an even greater degree of control.

You can install additional plugins simply by copying the Java class files to the plugins directory in the installation directory. If the plugin has been created with the required set of methods then Spam Annihilator will use it to process emails. Any variable data files that the plugin uses should be located in the data directory and any static files (files that do not change over time) should be located in the files directory.

Plugin Special Notes

Bayes
This plugin needs training before it will become effective. Due to memory requirements you may experience problems if you try to train with too many messages at a time. Training with less than 400 messages should present no problems but it really depends on your system and the amount of virtual memory available. If you experience problems then limit the number of emails you train with.

It is as important to train with legitimate emails as it is to train with spam. If you regularly get a good deal of spam then try to save it for training purposes.

BlockExe
You are encouraged to use this plugin. It is simple but very effective and it will practically eliminate email borne virus. This does NOT mean it is then safe to click on attachments that you are not expecting even if they appear to come from a friend or colleage. That is still a very stupid thing to do.

CharSet
Pay special attention to the warnings about only allowing certain character sets. If you are going to use this plugin then you are advised to deny certain character sets rather than only allow certain character sets.

Dictionary
This plugin is experimental and will probably lead to many false positives since legitimate emails often contain misspelled words. A dictionary has not been distributed with this plugin. To install a dictionary for this plugin create or obtain a file that is a list of words one per line and place it in the data directory of the installation directory as a file called Dictionary (case is important on case sensitive operating systems). On unix systems you may find a dictionary in /usr/share/dict and there are dictionaries for many languages at ftp://ftp.ox.ac.uk/pub/wordlists

If you are going to create a dictionary it is important to note that the file MUST be simple ascii text i.e. created with a program like Notepad on windows and not a word processor.

Garbage
This plugin is experimental. It may or may not lead to false positives. It looks for sequences of characters that do not occur in any word in a dictionary. It has been installed with a list of character tuples that do not occur in the English language. A character tuple is a combination of two characters. In fact this plugin will work with any length of character sequences but since the permutations would grow exponentially using longer character sequences will increase substantially the processing time for this plugin.

The plugin reads the tuples from a file located in the data directory called GarbageBadJuxta (case is important on case sensitive operating systems). This file contains a list of tuples one per line. To install a list of character tuples for another language create a file that contains a list of invalid character tuples for the chosen language, one per line and replace the existing file with the new file keeping the same name.

If you are going to create a new tuple file it is important to note that the file MUST be simple ascii text i.e. created with a program like Notepad on windows and not a word processor.

If you are going to create a new tuple file from a dictionary you should consider using a perl or shell script to generate all the permutations of tuples in the dictionary's character set and then check for their existence or lack thereof in the dictionary.

SpamHaus
This plugin is effective in trapping email that comes from systems that are used by spammers. The downside is that it will take longer to process each email because of the DNS (Domain Name Server) lookups. A request is sent to a DNS server on the internet and the plugin will wait for a reply. Without DNS lookups emails can be processed in as little as 50 milli seconds with the average being something like 400 ms (depending on the speed and memory of your system). With DNS lookups this could increase to 10 seconds or more if your internet connection is very slow. If you have a fast internet connection then you don't need to worry and you are encouraged to use this plugin module.