diff options
| author | Michael Denkowski <mdenkows@cs.cmu.edu> | 2014-02-26 12:15:02 -0800 |
|---|---|---|
| committer | Michael Denkowski <mdenkows@cs.cmu.edu> | 2014-02-26 12:15:02 -0800 |
| commit | 75fc827d0f6364c4e69f949abcbaf858f6735193 (patch) | |
| tree | 10cca0c435d74fd36406acda5184716f526862a2 /python | |
| parent | 73bc12bb80316362c3ba0c5a745bfc9591f88c55 (diff) | |
Comments
Diffstat (limited to 'python')
| -rw-r--r-- | python/cdec/sa/features.py | 3 | ||||
| -rw-r--r-- | python/cdec/sa/rulefactory.pxi | 1 |
2 files changed, 3 insertions, 1 deletions
diff --git a/python/cdec/sa/features.py b/python/cdec/sa/features.py index f5cbdb8d..dcc60401 100644 --- a/python/cdec/sa/features.py +++ b/python/cdec/sa/features.py @@ -147,6 +147,7 @@ def CountExceptLM(vocab): return CountExceptLM def CountExceptLex(ttable): - def CountExceptLex(ctx): # Word count in online data but NOT original bitext + 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 diff --git a/python/cdec/sa/rulefactory.pxi b/python/cdec/sa/rulefactory.pxi index 2be5cad6..78a23196 100644 --- a/python/cdec/sa/rulefactory.pxi +++ b/python/cdec/sa/rulefactory.pxi @@ -2052,6 +2052,7 @@ cdef class HieroCachingRuleFactory: stats.phrases_al[f_ph][e_ph] = al # Update Bilexical counts + # TODO: use alignments instead of cooc for e_w in e_words: stats.bilex_e[e_w] += 1 for f_w in f_words: |