-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathdata_process.py
164 lines (159 loc) · 6.31 KB
/
data_process.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
import math
import config
from helper import get_gram2id
def getBMES(input_str):
"""
将每个输入词转换为BMES标注
"""
output_str = []
if len(input_str) == 1:
output_str.append('S')
elif len(input_str) == 2:
output_str = ['B', 'S']
else:
M_num = len(input_str) - 2
M_list = ['I'] * M_num
output_str.append('B')
output_str.extend(M_list)
output_str.append('E')
return output_str
def load_data(data_dir):
"""
返回句子、BMES标签、词性标注标签
"""
with open(data_dir, 'r', encoding='utf-8-sig') as f:
word_list = []
seg_list = []
pos_list = []
segpos_list = []
flag =[] #用于标志前后句之间是否有关联，0：无关联；1：前后句之间有关联
num_conjun = 1 #用于标志句子间的联系
tail_pun=['。','”','！','？','；'] #语句结尾可能的标点符号
max_len = 511
k = 0
for line in f:
k += 1
if len(line)==1:
continue
word = []
seg = []
pos = []
segpos =[]
pretail = 0
temptail = 0
line=line.strip('\n')
line=line.strip(' ')
text=line.split(' ')
for word_pos in text:
sp_c = word_pos.split('/')
word_c = sp_c[0]
pos_c = sp_c[-1]
if pos_c == '。':
print(line)
length_c = len(word_c)
if length_c == 1:
word.append(word_c)
seg.append('S')
pos.append(pos_c)
segpos.append('S'+pos_c)
elif length_c == 2:
word.append(word_c[0])
seg.append('B')
pos.append(pos_c)
segpos.append('B' + pos_c)
word.append(word_c[1])
seg.append('E')
pos.append(pos_c)
segpos.append('E' + pos_c)
else:
if word_c:
word.append(word_c[0])
seg.append('B')
pos.append(pos_c)
segpos.append('B' + pos_c)
for middle_c in word_c[1:-1]:
word.append(middle_c)
seg.append('I')
pos.append(pos_c)
segpos.append('I' + pos_c)
word.append(word_c[-1])
seg.append('E')
pos.append(pos_c)
segpos.append('E' + pos_c)
else:
print(text)
print(k)
print('Sentence data error!')
length = len(word)
assert length == len(seg)
assert length == len(pos)
assert length == len(segpos)
if length > max_len:
prestart = 0
for i in range(0,length):
if (i==length-1) or ((word[i] in tail_pun) and (word[i+1] not in tail_pun)):
pretail = temptail
temptail = i
if pretail < prestart + max_len and temptail >= prestart + max_len:
word_list.append(['[CLS]']+word[prestart : pretail+1])
seg_list.append(seg[prestart : pretail+1])
pos_list.append(pos[prestart : pretail+1])
segpos_list.append(segpos[prestart: pretail+1])
prestart = pretail+1
flag.append(num_conjun)
if prestart <= length-1:
word_list.append(['[CLS]'] + word[prestart: ] )
seg_list.append( seg[prestart: ] )
pos_list.append( pos[prestart: ] )
segpos_list.append( segpos[prestart: ] )
flag.append(num_conjun)
num_conjun = num_conjun + 1
else:
word_list.append(['[CLS]'] + word )
seg_list.append( seg )
pos_list.append( pos)
segpos_list.append( segpos)
flag.append(0)
gram2id = []
gram2count = []
# gram2id, gram2count = get_gram2id(word_list, config.ngram_type, config.ngram_length, config.ngram_threshold)
gram_list = []
positions = []
gram_maxlen = []
for sentence in word_list:
sentence = sentence[1:]
if config.use_attention :
ngram_list = []
matching_position = []
ngram_list_len = []
for i in range(config.cat_num):
ngram_list.append([])
matching_position.append([])
ngram_list_len.append(0)
for i in range(len(sentence)):
for j in range(0, config.ngram_length):
if i + j + 1 > len(sentence):
break
ngram = ''.join(sentence[i: i + j + 1])
if ngram in gram2id:
index = int(min(config.cat_num, math.log2(gram2count[ngram]))) - 1
assert 0 <= index < config.cat_num
channel_index = index
try:
index = ngram_list[channel_index].index(ngram)
except ValueError:
ngram_list[channel_index].append(ngram)
index = len(ngram_list[channel_index]) - 1
ngram_list_len[channel_index] += 1
for k in range(j + 1):
matching_position[channel_index].append((i + k, index))
else:
ngram_list = None
matching_position = None
ngram_list_len = None
max_ngram_len = max(ngram_list_len) if ngram_list_len is not None else None
gram_list.append(ngram_list)
positions.append(matching_position)
gram_maxlen.append(max_ngram_len)
return word_list,seg_list,pos_list,segpos_list,flag,gram_list,positions,gram_maxlen,gram2id
# sentences,seg,pos,segpos,flag=load_data('zuozhuan_train_utf8.txt')