db maintenance .MSG_COUNT
Jef Poskanzer
jef at acme.com
Fri Mar 5 16:54:03 CET 2004
How about scaling it? Count up the total number of token-instances
in the db before and after maintenance, and make a new .MSG_COUNT
with the same tokens/message ratios as before.
For example, let's say I had this db:
.MSG_COUNT 10 10 20040229
viagra 10 0 20040229
jef 10 10 20040229
xyzzy 0 5 20031201
plugh 0 5 20031201
Total number of token-instances is 20 spam and 20 ham. Then I remove the
tokens from 2003, leaving this:
viagra 10 0 20040229
jef 10 10 20040229
Now the total number of token-instances is 20 spam and only 10 ham,
therefore .MSG_COUNT should be scaled to 10 5.
More information about the Bogofilter
mailing list