diff options
Diffstat (limited to 'python/cdec/sa/features.py')
| -rw-r--r-- | python/cdec/sa/features.py | 11 |
1 files changed, 11 insertions, 0 deletions
diff --git a/python/cdec/sa/features.py b/python/cdec/sa/features.py index c8fc1cca..dcc60401 100644 --- a/python/cdec/sa/features.py +++ b/python/cdec/sa/features.py @@ -140,3 +140,14 @@ def IsSupportedOnline(ctx): # Occurs in online data? return (ctx.online.paircount > 0.01) else: return False + +def CountExceptLM(vocab): + def CountExceptLM(ctx): # Word count in bitext (inc online data) but NOT mono text + return sum(1 for e in ctx.ephrase.words if e not in vocab) + return CountExceptLM + +def CountExceptLex(ttable): + def CountExceptLex(ctx): # Word count in online data but NOT aligned in original bitext + # TODO: Check that online data actually contains aligned word when rulefactory TODO is addressed. + return sum(1 for e in ctx.ephrase.words if not ttable.contains_e_word(e)) + return CountExceptLex |