summaryrefslogtreecommitdiff
path: root/python/src
diff options
context:
space:
mode:
authorMichael Denkowski <michael.j.denkowski@gmail.com>2013-01-24 16:00:14 -0500
committerMichael Denkowski <michael.j.denkowski@gmail.com>2013-01-24 16:00:14 -0500
commitaa99dfbcd324a23814fbaea3ac22d42bb0289000 (patch)
tree11d54c3b27806156ff717f29afd1ba53e69f552d /python/src
parent2ade6dc6983a5c26306f5d3ecb862d50df06b024 (diff)
Clean up dead code
Diffstat (limited to 'python/src')
-rw-r--r--python/src/sa/default_scorer.pxi74
1 files changed, 0 insertions, 74 deletions
diff --git a/python/src/sa/default_scorer.pxi b/python/src/sa/default_scorer.pxi
deleted file mode 100644
index 483f4743..00000000
--- a/python/src/sa/default_scorer.pxi
+++ /dev/null
@@ -1,74 +0,0 @@
-from libc.stdlib cimport malloc, realloc, free
-from libc.math cimport log10
-
-MAXSCORE = -99
-EgivenFCoherent = 0
-SampleCountF = 1
-CountEF = 2
-MaxLexFgivenE = 3
-MaxLexEgivenF = 4
-IsSingletonF = 5
-IsSingletonFE = 6
-NFEATURES = 7
-
-cdef class DefaultScorer(Scorer):
- cdef BiLex ttable
- cdef int* fid
-
- def __dealloc__(self):
- free(self.fid)
-
- def __init__(self, BiLex ttable):
- self.ttable = ttable
- self.fid = <int*> malloc(NFEATURES*sizeof(int))
- cdef unsigned i
- for i, fnames in enumerate(('EgivenFCoherent', 'SampleCountF', 'CountEF',
- 'MaxLexFgivenE', 'MaxLexEgivenF', 'IsSingletonF', 'IsSingletonFE')):
- self.fid[i] = FD.index(fnames)
-
- cdef FeatureVector score(self, Phrase fphrase, Phrase ephrase,
- unsigned paircount, unsigned fcount, unsigned fsample_count):
- cdef FeatureVector scores = FeatureVector()
-
- # EgivenFCoherent
- cdef float efc = <float>paircount/fsample_count
- scores.set(self.fid[EgivenFCoherent], -log10(efc) if efc > 0 else MAXSCORE)
-
- # SampleCountF
- scores.set(self.fid[SampleCountF], log10(1 + fsample_count))
-
- # CountEF
- scores.set(self.fid[CountEF], log10(1 + paircount))
-
- # MaxLexFgivenE TODO typify
- ewords = ephrase.words
- ewords.append('NULL')
- cdef float mlfe = 0, max_score = -1
- for f in fphrase.words:
- for e in ewords:
- score = self.ttable.get_score(f, e, 1)
- if score > max_score:
- max_score = score
- mlfe += -log10(max_score) if max_score > 0 else MAXSCORE
- scores.set(self.fid[MaxLexFgivenE], mlfe)
-
- # MaxLexEgivenF TODO same
- fwords = fphrase.words
- fwords.append('NULL')
- cdef float mlef = 0
- max_score = -1
- for e in ephrase.words:
- for f in fwords:
- score = self.ttable.get_score(f, e, 0)
- if score > max_score:
- max_score = score
- mlef += -log10(max_score) if max_score > 0 else MAXSCORE
- scores.set(self.fid[MaxLexEgivenF], mlef)
-
- # IsSingletonF
- scores.set(self.fid[IsSingletonF], (fcount == 1))
-
- # IsSingletonFE
- scores.set(self.fid[IsSingletonFE], (paircount == 1))
-
- return scores