-
Notifications
You must be signed in to change notification settings - Fork 0
/
rouge.py
164 lines (143 loc) · 6.18 KB
/
rouge.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
from nltk.util import ngrams
from nltk.tokenize import TreebankWordTokenizer, PunktSentenceTokenizer
from nltk import skipgrams
import numpy as np
tokenizer = TreebankWordTokenizer()
sentence_tokenizer = PunktSentenceTokenizer()
class Rouge():
def jacknifing(score_list, averaging=True):
if(len(score_list) == 1):
return(np.mean(score_list))
elif((len(score_list) > 1) and (averaging is False)):
return(score_list)
else:
for i in range(len(score_list)):
average = []
dummy = list(score_list)
dummy.remove(dummy[i])
average.append(max(dummy))
return(np.mean(average))
def get_score(r_lcs, p_lcs, beta):
try:
score = ((1+beta**2)*r_lcs*p_lcs)/(r_lcs+(beta**2)*p_lcs)
except ZeroDivisionError as e:
score = 0
return score
def lcs(X, Y, m, n):
L = [[0 for x in range(n+1)] for x in range(m+1)]
for i in range(m+1):
for j in range(n+1):
if i == 0 or j == 0:
L[i][j] = 0
elif X[i-1] == Y[j-1]:
L[i][j] = L[i-1][j-1] + 1
else:
L[i][j] = max(L[i-1][j], L[i][j-1])
index = L[m][n]
lcs = [""] * (index+1)
lcs[index] = ""
i = m
j = n
while i > 0 and j > 0:
if X[i-1] == Y[j-1]:
lcs[index-1] = X[i-1]
i -= 1
j -= 1
index -= 1
elif L[i-1][j] > L[i][j-1]:
i -= 1
else:
j -= 1
s = " ".join(lcs)
return(len(s.split()), s)
def rouge_n(references, candidate, n, averaging=True):
ngram_cand = ngrams(tokenizer.tokenize(candidate), n)
ng_cand = list(ngram_cand)
rouge_recall = []
for ref in references:
count = 0
ngram_ref = ngrams(tokenizer.tokenize(ref), n)
ng_ref = list(ngram_ref)
for ngr in ng_cand:
if ngr in ng_ref:
count += 1
rouge_recall.append(count/len(ng_ref))
return Rouge.jacknifing(rouge_recall, averaging=averaging)
def rouge_l_sentence(references, candidate, beta, averaging=True):
rouge_l_list = []
for ref in references:
arg1 = tokenizer.tokenize(ref)
arg2 = tokenizer.tokenize(candidate)
r_lcs = Rouge.lcs(arg1, arg2, len(arg1), len(arg2))[0]/len(arg1)
p_lcs = Rouge.lcs(arg1, arg2, len(arg1), len(arg2))[0]/len(arg2)
score = Rouge.get_score(r_lcs, p_lcs, beta=beta)
rouge_l_list.append(score)
return Rouge.jacknifing(rouge_l_list, averaging=averaging)
def rouge_l_summary(references, candidate, beta, averaging=True):
rouge_l_list = []
cand_sent_list = sentence_tokenizer.tokenize(candidate)
for ref in references:
ref_sent_list = sentence_tokenizer.tokenize(ref)
sum_value = 0
for ref_sent in ref_sent_list:
l_ = []
arg1 = tokenizer.tokenize(ref_sent)
for cand_sent in cand_sent_list:
arg2 = tokenizer.tokenize(cand_sent)
d = tokenizer.tokenize(Rouge.lcs(arg1, arg2,
len(arg1), len(arg2))[1])
l_ += d
sum_value = sum_value+len(np.unique(l_))
r_lcs = sum_value/len(tokenizer.tokenize(ref))
p_lcs = sum_value/len(tokenizer.tokenize(candidate))
score = Rouge.get_score(r_lcs, p_lcs, beta=beta)
rouge_l_list.append(score)
return Rouge.jacknifing(rouge_l_list, averaging=averaging)
def normalized_pairwise_lcs(references, candidate, beta, averaging=True):
normalized_list = []
cand_sent_list = sentence_tokenizer.tokenize(candidate)
for ref in references:
ref_sent_list = sentence_tokenizer.tokenize(ref)
scr = []
for r_sent in ref_sent_list:
s = []
arg1 = tokenizer.tokenize(r_sent)
for c_sent in cand_sent_list:
arg2 = tokenizer.tokenize(c_sent)
s.append(Rouge.lcs(arg1, arg2, len(arg1), len(arg2))[0])
scr.append(max(s))
r_lcs = 2*sum(scr)/len(tokenizer.tokenize(ref))
p_lcs = 2*sum(scr)/len(tokenizer.tokenize(candidate))
score = Rouge.get_score(r_lcs, p_lcs, beta=beta)
normalized_list.append(score)
return Rouge.jacknifing(normalized_list, averaging=averaging)
def rouge_s(references, candidate, beta, d_skip=None, averaging=True, smoothing=False):
rouge_s_list = []
k_c = len(candidate) if d_skip is None else d_skip
cand_skip_list = list(skipgrams(tokenizer.tokenize(candidate),
n=2, k=k_c))
for ref in references:
k_ref = len(ref) if d_skip is None else d_skip
ref_skip_list = list(skipgrams(tokenizer.tokenize(ref),
n=2, k=k_ref))
count = 0
for bigram in cand_skip_list:
if bigram in ref_skip_list:
count = count+1
if not smoothing:
r_skip = count/len(ref_skip_list)
p_skip = count/len(cand_skip_list)
else:
cand_ungm = list(ngrams(tokenizer.tokenize(candidate),
n=1))
ref_ungm = list(ngrams(tokenizer.tokenize(ref),
n=1))
for ungm in cand_ungm:
if ungm in ref_ungm:
count += 1
r_skip = count/(len(ref_skip_list)+len(ref_ungm))
p_skip = count/(len(cand_skip_list)+len(cand_ungm))
score = Rouge.get_score(r_skip, p_skip, beta)
rouge_s_list.append(score)
return Rouge.jacknifing(rouge_s_list, averaging=averaging)
print(Rouge.rouge_s(['police killed the gunman'],'gunman the killed police',beta=1,averaging=False,smoothing=True))