Humanist Discussion Group, Vol. 17, No. 95.
Centre for Computing in the Humanities, King's College London
www.kcl.ac.uk/humanities/cch/humanist/
Submit to: humanist@princeton.edu
Date: Tue, 17 Jun 2003 06:40:08 +0100
From: Virginia Knight <Virginia.Knight@bristol.ac.uk>
Subject: Re: 17.086 anti-spam
--On 14 June 2003 08:51 +0100 "Humanist Discussion Group (by way of Willard
McCarty <willard.mccarty@kcl.ac.uk>)" <willard@lists.village.virginia.edu>
wrote:
>SpamAssassin appears to be working very well for me. So far in ca 2 weeks
>of using it, spam has been reduced by at least 90% and no message I have
>wanted to read has been wrongly classified. I can certainly live with the
>10% while the mechanism for filtering is improved. It will be interesting
>to see if spammers continue to learn from the techniques used against
>them. Here, for example, is a typical analysis of a spamming message:
I was given a peep at some of the code for just one of the SpamAssassin
rules and it proved to be pretty complicated. This was in the course of
reporting a bug (since fixed). I was finding that reviews from an
e-journal were being flagged as spam because the abbreviation 'pp' was
occurring near words such as 'longer'!
If you run a SpamAssassin installation you can tweak it to reweight the
various tests it performs. Failing that, you can still customise it by
altering the points threshold for spam, or (as I do) combining SpamAssassin
with a procmail filter which automatically sends a message which tests
positive on certain SpamAssassin tests to a spam folder. This is because
spam often falls below even a low points threshold. I find for example
that treating messages with a large amount of HTML in them as spam catches
some spam that SpamAssassin misses, at the price of treating as spam some
genuine messages which come from people who aren't regular correspondents
of mine. I know others have a similar rule for filtering email, so maybe
people who compose email in HTML should note that it is risky to use a
format so associated with spammers.
Virginia Knight
----------------------
Virginia Knight, Institute for Learning and Research Technology
Tel: +44 (0)117 928 7154 Fax: +44 (0)117 928 7112
University of Bristol, 8-10 Berkeley Square, Bristol BS8 1HH
Virginia.Knight@bristol.ac.uk
Official homepage: http://www.ilrt.bris.ac.uk/aboutus/staff?search=cmvhk
Personal homepage: http://www.ilrt.bris.ac.uk/~ggvhk/virginia.html
ILRT homepage: http://www.ilrt.bristol.ac.uk
This archive was generated by hypermail 2b30 : Tue Jun 17 2003 - 01:49:39 EDT