diff options
| author | Michael Denkowski <mdenkows@cs.cmu.edu> | 2014-03-07 13:21:40 -0800 |
|---|---|---|
| committer | Michael Denkowski <mdenkows@cs.cmu.edu> | 2014-03-07 13:21:40 -0800 |
| commit | f0e60d27a546c2cc429bb04e16c373e2aecd9ed2 (patch) | |
| tree | 93fb5b9ffe51ee7cd39f07d0de9cc4e9eb558e38 /python/cdec/sa | |
| parent | abcd6865a25944a1cd07c9224db2fd7a729f02e6 (diff) | |
Unsmoothed bilex. Cython uses smoothing?
Diffstat (limited to 'python/cdec/sa')
| -rw-r--r-- | python/cdec/sa/online.py | 25 |
1 files changed, 14 insertions, 11 deletions
diff --git a/python/cdec/sa/online.py b/python/cdec/sa/online.py index 008793d8..0783e727 100644 --- a/python/cdec/sa/online.py +++ b/python/cdec/sa/online.py @@ -36,9 +36,10 @@ class Bilex: def get_score(self, f, e, dir): if dir == 0: - return self.p_fe(f, e) - if dir == 1: - return self.p_ef(e, f) + p = self.p_fe(f, e) + elif dir == 1: + p = self.p_ef(e, f) + return p def p_fe(self, f, e): d = self.fe.get(f, None) @@ -60,22 +61,22 @@ class Bilex: aligned_fe[i].append(j) aligned_ef[j].append(i) for f_i in range(len(f_words)): - self.f[f_words[f_i]] += 1 e_i_aligned = aligned_fe[f_i] - lc = len(e_i_aligned) - if lc > 0: + if len(e_i_aligned) > 0: for e_i in e_i_aligned: - self.fe[f_words[f_i]][e_words[e_i]] += (1 / lc) + self.f[f_words[f_i]] += 1 + self.fe[f_words[f_i]][e_words[e_i]] += 1 else: + self.f[f_words[f_i]] += 1 self.fe[f_words[f_i]][NULL_WORD] += 1 for e_i in range(len(e_words)): - self.e[e_words[e_i]] += 1 f_i_aligned = aligned_ef[e_i] - lc = len(f_i_aligned) - if lc > 0: + if len(f_i_aligned) > 0: for f_i in f_i_aligned: - self.ef[e_words[e_i]][f_words[f_i]] += (1 / lc) + self.e[e_words[e_i]] += 1 + self.ef[e_words[e_i]][f_words[f_i]] += 1 else: + self.e[e_words[e_i]] += 1 self.ef[e_words[e_i]][NULL_WORD] += 1 # Update counts from alignd bitext @@ -103,9 +104,11 @@ class Bilex: for f in fv: for (e, c) in sorted(self.fe[f].iteritems()): out.write('{} {} {}\n'.format(f, e, c)) + out.write('\n') for e in ev: for (f, c) in sorted(self.ef[e].iteritems()): out.write('{} {} {}\n'.format(e, f, c)) + out.write('\n') def read(self, in_f): with gzip_or_text(in_f) as inp: |