[wordup] E-texts used against Bayesian spam-filters

Adam Shand adam at shand.net
Fri Dec 5 05:21:50 EST 2003


I was wondering how long it would take for them to figure out how to 
subvert bayesian filters.  Sadly I think the deal is just that email is 
dying.  It may takes years and years and don't know what, if anything, 
will replace it but email is dying.  The bottom line is that spammers 
have more to gain then any individual has to lose from spam.  Economics 
is against us.

Adam.

From: http://boingboing.net/2003_12_01_archive.html#107036645218560926
More: http://news.bbc.co.uk/1/hi/technology/3247200.stm

E-texts used against Bayesian spam-filters

Bayesian anti-spam filters count word-frequency in suspect messages and 
compare the results to profiles of word-frequency in spam and ham. 
Defeating this requires that your spam include a lot of natural human 
prose. So spammers have started to mine the Gutenberg Project and other 
sources of human-generated ASCII and dumping random hunks of literature 
into their messages to get around the filters.
Blogger and journalist Clive Thompson found an excerpt from Chapter 20 
of The Master Key by Wizard of Oz author L Frank Baum in a message that 
had as its subject line "the big unit" (no prizes for guessing what the 
rest of it was hawking).
Link




More information about the wordup mailing list