NAME¶
slmprune - prune the back-off language model to a reasonable size
SYNOPSIS¶
slmprune input_slm result_slm cut_option
num...
DESCRIPTION¶
This program uses entropy-based method to prune the size of back-off language
model '
input_slm' to a specific size and write to '
result_slm'.
the third parameter
cut_option can be [
R|
C]. It means the
following numbers is the number for (
R)eserve or (
C)ut. If
(
C)ut, the num[k] means how many items in level K would be cut. If (
R)eserve, num[k] means how many item would be reserved in level k.
Note that we do not ensure that during pruning process, exactly the the given
number of items are cut or reserved, because some items may contains high
level children, so could not be cut.
Also it's your responsiblity to give right number of arguments based on
'input_slm'.
NOTE¶
To get information of the back-off language model, try '
slminfo'.
AUTHOR¶
Originally written by Phill.Zhang <phill.zhang@sun.com>. Currently
maintained by Kov.Chai <tchaikov@gmail.com>.
SEE ALSO¶
slminfo(1),
slmthread(1).