The anatomy of a spam

The anatomy of a spam according to the BBC. It’s quite interesting when you read the article.

The fact that the emailaddress is spoofed and that spammers use real domain names is nothing significant new. What makes it more difficult today is the large amount of text at the end or even te beginning of the email. This text block is full of good words and is there to corrupt your Bayesian database when you try to index the content. So be carefull if you manage your anti spam solution.

The included image, that’s why we call it image based spam, is more tricky. Some vendors and anti spam providers now use OCR to detect the content inside the image. Based on this they will mark, quarantine or reject the spam.

Since a few weeks now, I see that spammers are already implementing new techniques for the image based spam. They will make the image more difficult to read for filters that use OCR technologies. For example by using more background colors and patterns, placing strokes in different colors over the text and even placing all the letters with different spacing and not on the traditional horizontal line.

At MX Lab I also use a tools that uses OCR and some other technologies to block this type of spam but now we have created a more explicit filter to intercept this kind of spam and with very good results.

Read the full article at their site.