The Nigerian Spammer Anthem

A recent Los Angeles Times article reveals that the 419 scam spammers have their very own anthem: a song titled I Go Chop Your Dollars by Nigerian recording artist Osofia:


This is a companion discussion topic for the original blog entry at: http://www.codinghorror.com/blog/2005/10/the-nigerian-spammer-anthem.html

SpamRIP is adware. Better stay away.

I am very close to 99.5% with my Bayesian. Take a look at the statistics:

http://www.newsforge.com.ar/images/Spamsieve3.png

There have been some negatives here and there (and false positives) but so far, I haven’t read a spam in months! I completely forgot about its existance. :slight_smile:

I recently ditched my long standing e-mail address, simply because of the amount of spam I recieved.

I did use POPFile, but it was the fact that every morning when I turned my computer on, it had to download around 1000 e-mails a day, with 99.9% being spam.

These days I’ve got two e-mails address, one super secret one just for family and friends. The other for using in places that are likely to get it on a spam list.

Once again, something new. I’ve heard of approaches using Bayesian with two word and three word phrases, but I’ve never heard of Markovian approach.

Is the approach similar to Bayesian in that it uses statistical inference, but just with phrases?

I am very close to 99.5% with my Bayesian.

The Markovian filtering would deliver something like 99.99% in this case… rule of thumb is, it cuts the amount of spam that makes it through a bayesian filter in half.

Is the approach similar to Bayesian in that it uses statistical inference, but just with phrases?

Yes. From page 20 of the slide deck (which I still highly recommend – it’s great)

http://crm114.sourceforge.net/Plateau99.pdf

How to Turn a Bayesian into a Markovian

(1) change the feature generator from single words to spanning multiple words *

(2) Change the weighting so that longer features have more weight (ie. Longer features generate local probabilities closer to 0.0 and 1.0)

(3) The 22n weighting means that the
weights were 1, 4, 16, 64, 256, … for span
lengths of 1, 2, 3, 4, 5 … words

Wow-- here’s the music video for “I Go Chop Your Dollars”:

http://nigeriamovies.net/419.htm

Pretty funny. It’s a lot less gangsta rap than I was expecting, with scenes of the mark signing contracts, touring Nigeria, etc.