as largely as Hakeem renders, you can tell the weaver much more mainly

keywords that match against the routing
information in the email header: who it came from, who it is going to.
The phone analogy is recipient and originating phone numbers.
This cuts down on "noise". "Secret Power" gave examples of this too.

o Search all traffic for a set of keywords that are found (tuned) over
time to have the best results. SOME of the ones I used:

first day
last day
resign
new job
resume
interview
drug test

It's ironic that drug testing of employees is so wide-spread that it can
be used to pick out people looking for new jobs.

o Further exclusion logic (keywords) to isolate the meaning of the
keyword 'resume' to mean job history. Also, UK people say 'CV'.
Example: do NOT allow a sentence fragment like 'resume playing'
to trigger "resume condition" inclusion.

That's how it is done.

I then sit at a terminal and page through a
summary of the results, looking for 'hits'.

That's how DICTIONARY works too.

* The Puzzle Palace, Author James Bamford, 1983 revision
*
* P496-497: You would put in a whole slew of keywords.
* You flip through the results.

And it's damn effective.

I could pick needles out of a haystack. I could find a 16-line Risk Management
report in Salomon's daily 150-230 megabytes of Internet email traffic. It took
only one word: 'risk', and lots of exclusion logic, because the word is used
lots. I had never seen that format of risk report before. It was incoming too.

It sure didn't look like much, but...

The head of Risk M