Dealing with previously spammed IP addresses in Anti-Splog

On the 23:d march the following users where marked as spam with 100%
certainty (I had to edit out the addresses on this forum, so please pm me or something so I can send you the actual addresses in order to look at your logs);

x@x.com
x@x.com
x@x.com

None of them where spam, but all came from the same IP. At the same
time;

x@x.com
x@x.com
x@x.com

Also registered, but I see no trace at all at my blog. Maybe you
aromatically just refuses everything from one IP when you have decided
its spam, I dont know.

Anyway, 100% seems not like a good figure. All of the above addresses
are government addresses here in Sweden. Ok, they might have typed a
"no-no" word, maybe have a miss-configured e-mail server, but it cant
be 100%. 100% means that you are SURE its spam, thats a big difference
from you expect its probably spam.

Now my users starts to ask why so many things are caught as false
positives, what am I going to say? My users are either teachers or
students. They all register in hordes from the same IP during class
hour, so when its wrong many users and the class time suffer.

Will you incorporate some sort of fine-gradient score system or
should I expect the score mechanism to stay the same?