def BLEU(hyp, ref):
# print 'BLEU'
metric = 0.0
hyp_list = split_by_words_and_punctuation(hyp.lower())
ref_list = split_by_words_and_punctuation(ref.lower())
hyp_len = len(hyp_list)
ref_len = len(ref_list)
h_list = split_into_ngramm(hyp, 5)
r_list = split_into_ngramm(ref, 5)
if ref_len < hyp_len:
Bp = 1
else:
Bp = exp(1 - (float(hyp_len) / ref_len))
sum = 0.0
N = 4
for i in range(1, N + 1):
try:
p = log(Pn(h_list, r_list, i), 2)
except:
p = -99999
sum += (p / float(N))
metric = Bp * exp(sum)
print 'BLEU =', metric
return metric
def BLEU(hyp, ref):
# print 'BLEU'
metric = 0.0
hyp_list = split_by_words_and_punctuation(hyp.lower())
ref_list = split_by_words_and_punctuation(ref.lower())
hyp_len = len(hyp_list)
ref_len = len(ref_list)
h_list = split_into_ngramm(hyp, 5)
r_list = split_into_ngramm(ref, 5)
if ref_len < hyp_len:
Bp = 1
else:
Bp = exp(1-(float(hyp_len)/ref_len))
sum = 0.0
N = 4
for i in range(1, N+1):
try:
p = log(Pn(h_list, r_list, i), 2)
except:
p = -99999
sum += (p/float(N))
metric = Bp*exp(sum)
print 'BLEU =', metric
return metric
def bitext(orig, trans):
for (o, t) in zip(orig, trans):
yield (split_into_ngramm(o, 5), split_into_ngramm(t, 5))
def bitext(orig, trans):
for (o, t) in zip(orig, trans):
yield (split_into_ngramm(o, 5), split_into_ngramm(t, 5))
