Spam happens. No matter how hard you try, eventually your email address(es) will get spammed by something wanting to sell you medication or receive funds from a Nigerian prince. Aside from proactively keeping your email secret from the world, the only defense against these emails is filters, but like any defense, there are ways around them. Now researchers at Concordia University are working on a new kind of filter that will better recognize spam designed to defeat modern filters.
Just like malware authors, spammers can be clever people that study defenses to find weaknesses. For example, by replacing letters with symbols, a text-focused filter will fail to stop the email, and image-based filters can be fooled by similar images with different backgrounds and colors. If spam consisted of both text and images, it could completely defeat almost every filter out there, which is why the researchers are looking to create a new filtering technique. This new technique will employ data mining methods for identifying patterns across text and images, so the combination will not work.
This new filtering technique is still in development at the moment, but researchers are working to release a plug-in for SpamAssassin, the world's most popular open source spam filter. Hopefully they will finish before spammers find a way to defeat this defense too.