What to Filter?

One of the hardest parts of managing a spam filter application is figuring out what to filter and what not to.

By using a fuzzy logic filter to stop junk email, the decision doesn't have to be black and white. Items that frequently appear in spam, but also occur in real email can still be factored in. Each individual filter has a "weight" to it which dictates how strong and indicator it is.

I recently discovered spammers were using a special table tag to reformat message text. After stripping out all the formatting, the message basically looks like a crossword puzzle. This makes the content itself near impossible to process. Fortunately, the table tag itself is easy to detect. Some research indicated that it pretty much is never used in legitimate emails. The offending tag is now in the antispam filter as a very "strong" rule.

The vast majority of spam is loaded with junk HTML. They usually include tables, images and numerous font tags. Unfortunately, some mail clients like Outlook can also really load messages up with trash.

Incrapomail (you know who I'm talking about) is truly the kind of crap content. Every message is loaded to the hilt with unneeded formatting. Then, users add their own junk. Typical messages include pictures of people's dogs, kids, cars and other stuff which doesn't have anything to do with anything.

My favorite emails include some sort of emoticon to help communicate the sender's pleasure or frustration with me.

The product in fact makes efforts to not to block email like this. I wish I could, but unfortunately a lot of users use that kind of junk-ware. Although, I sometimes wonder if the world would be better off if some of those messages just kind of "disappeared."

Back