A few months ago, my email address leaked out of its safe, comfortable residence in the Private Zone and into the world where people do bad things with your address. Each day I have several new spam in my mailbox, and what has been a problem for many people for quite some time has now come to roost on my doorstep. Perhaps that's not such a good analogy, but it's the best I can do right now.
Fortunately, I have read a lot about spam and email classification, mostly on the incredibly erudite articles written by Paul Graham. I find pattern recognition in general and email classification in particular to be fascinating topics, and now fate has provided me with a real-world, pertinent problem to solve. It's great how life works sometimes.
Many people are much further along the spam classification curve, and there are many software packages that can help. I have picked PopFile because it looks cool, and it seems to be working OK. I am in the process of "training" it to recognize spam, and I'll let you know how it progresses. For you techies, this is an email filter that uses Bayesian word-occurrence probability to classify email into categories. It gets its probabilities from the set of email that you receive, which is what makes it really cool - it learns to recognize your email over time.
Setup was quite easy, though my situation is (of course) a bit special. Let's just say that my email arrive like this: POP --(ssl)--> stunnel --> popfile --> pocomail. Privacy matters, remember.
6:02:20 PM 
|
|
