Newbie questions re registration

[Geoff]

Hello,

I am a new user and I am trying to understand the
registration process.  My MUA is sylpheed, which stores
mails in folders in which it simply numbers them from 1
upwards, and renumbers to extend the sequence or to
fill gaps when mail is added or deleted.

I have been using a simple script to train bogofilter on
these folders by sending mails to it one at a time
with the appropriate "s" or "n" option.  This appears to
work, but it means that, in my single wordlist.db I already
have several mails, both spam and ham, which were registered
under the identical names "1", "2" etc. My plan for the
future was to use the same system for new mails.  This would
obviously mean that the number of duplicate names
will grow.

On looking at the manpage I noticed the"S"
options to correct the wordlist, and this made me wonder how
bogofilter identifies registered mails internally?  I do not
know how, under the system I am using, I could identify a
mail for correction by the use of this option - which of my
numerous mails registered as "1" would be identified. This
makes me wonder if my whole approach is flawed because of
the duplicate names it produces, so that even my existing
wordlist is in some sense corrupt or inaccurate?

TIA

Geoff