That's very interesting!  I wonder, though, if you could re-run it
without any quoted emails?  I suspect that if you ran it against the
email as-is, you'd get 5 of any particular phrase in a thread just due
to quoting the same email in several other email.

Would be fun to run some statistical analysis against the whole
piclist archive.  I'd like to get a google trends like interface so I
can see the rise and fall of each PIC type, particular problems, and
other topics...

-Adam

On 4/14/08, Martin <martin@nnytech.net> wrote:
> I've been indexing piclist emails for the past few weeks. Here are the
> top 100 most popular phrases according to my scoring algorithm. It's
> "sort of" interesting. I don't know what I expected. Maybe when there
> are several thousand emails in the database it will look more
> interesting. A "phrase" here is defined as three unlikely words placed
> near each other. Often-occurring words such as "and the I is" etc. are
> not included for obvious reasons. The highest scoring phrases are mostly
> gibberish like: "psychotherapists psychopaths homeopaths" and "cpi ctaf
> displayitem" or "zhmezhme hansjurgen fajs".
-- 
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist