diff options
| author | Peng Wu <alexepico@gmail.com> | 2013-02-08 11:23:28 +0800 |
|---|---|---|
| committer | Peng Wu <alexepico@gmail.com> | 2013-02-08 11:33:46 +0800 |
| commit | c141d3e1e827d2fba968e83e55dfbfbedc967c1b (patch) | |
| tree | cb24dc656cb88e7474505836bc046963d3492f84 | |
| parent | a9ad1303f4235a8d147ff64f9bd2c8ed49a29a36 (diff) | |
| download | trainer-c141d3e1e827d2fba968e83e55dfbfbedc967c1b.tar.gz trainer-c141d3e1e827d2fba968e83e55dfbfbedc967c1b.tar.xz trainer-c141d3e1e827d2fba968e83e55dfbfbedc967c1b.zip | |
update config
| -rw-r--r-- | lib/myconfig.py | 4 | ||||
| -rw-r--r-- | tools/merge.py | 3 |
2 files changed, 4 insertions, 3 deletions
diff --git a/lib/myconfig.py b/lib/myconfig.py index 761aecd..d5ac9c4 100644 --- a/lib/myconfig.py +++ b/lib/myconfig.py @@ -142,10 +142,10 @@ class MyConfig: return 9 # minimum word occurrence in n-gram table def getPartialWordThreshold(self): - return 0.70 # the first 70% in position + return 0.60 # the first 60% in position def getNewWordThreshold(self): - return 0.70 # the first 70% in position, subject verb object. + return 0.79 # the first 79% in position, subject verb object. def getMinimumEntropy(self): return 0.01 diff --git a/tools/merge.py b/tools/merge.py index aa3b05d..ace5f41 100644 --- a/tools/merge.py +++ b/tools/merge.py @@ -9,7 +9,8 @@ threshold = 3 #minimum pinyin frequency #keep the un-merged word/pinyin/freq un-touched, #only affect the merged word/pinyin/freq tuples. -minimum = 3 +#try to use 2 to maximize the phrases in merged table. +minimum = 2 #default pinyin total frequency default = 100 |