-
Notifications
You must be signed in to change notification settings - Fork 7
/
utils_init.py
246 lines (224 loc) · 8.92 KB
/
utils_init.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
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
#!/usr/bin/env python
#-*- coding: utf-8 -*-
import random
import numpy as np
def load_vocab(filename):
vocab = []
with open(filename,encoding='utf-8',mode='r') as f:
for line in f:
vocab.append(line.strip())
vocab_to_id = {u:i for i,u in enumerate(vocab)}
return vocab,vocab_to_id
def load_pretrain(glove_file,word_emb_dim):
embedding_matrix,vocab = [], []
with open(glove_file,encoding='utf-8',mode='r') as f:
for i,line in enumerate(f):
if i % 1e5 == 0:
print('Current index is %d' %i)
try:
line_split = line.strip().split()
if len(line_split) == word_emb_dim + 1:
# if line_split[0] in vocab_set:
vocab.append(line_split[0])
embedding_matrix.append([float(x) for x in line_split[1:]])
except:
pass
vocab_to_id = {u:i for i,u in enumerate(vocab)}
id_to_vocab = {v:u for u,v in vocab_to_id.items()}
return embedding_matrix,vocab,vocab_to_id,id_to_vocab
def load_ED_data(filename,lower_case=False):
"""
loading ner data, sentence and its corresponding word-level ner label
"""
sents_all = []
ners_all = []
ner_1 = []
ner_2 = []
sent_tmp = []
ner_tmp = []
ner_1_tmp = []
ner_2_tmp = []
ner_vocab = set()
doc_file_to_sents = {}
with open(filename,encoding='utf-8',mode='r') as f:
w_last = ''
for line in f:
line = line.strip()
line_split = line.split(' ')
if len(line_split) == 5:
doc_file = line_split[1]
if lower_case:
line_split[0] = str(line_split).lower()
sent_tmp.append(line_split[0])
ner_tmp.append(line_split[-1])
ner_vocab.add(line_split[-1])
ner_1_tmp_tmp = line_split[2]
ner_1_tmp_tmp = ner_1_tmp_tmp
ner_1_tmp.append(ner_1_tmp_tmp)
ner_2_tmp_tmp = line_split[3]
ner_2_tmp_tmp = ner_2_tmp_tmp
ner_2_tmp.append(ner_2_tmp_tmp)
else:
if len(sent_tmp):
sents_all.append(sent_tmp)
ners_all.append(ner_tmp)
ner_1.append(ner_1_tmp)
ner_2.append(ner_2_tmp)
sent_tmp = []
ner_tmp = []
ner_1_tmp = []
ner_2_tmp = []
if doc_file not in doc_file_to_sents:
doc_file_to_sents[doc_file] = [len(sents_all) - 1]
else:
doc_file_to_sents[doc_file] += [len(sents_all) - 1]
w_last = line_split[0]
if len(sent_tmp) > 0:
sents_all.append(sent_tmp)
ners_all.append(ner_tmp)
ner_1.append(ner_1_tmp)
ner_2.append(ner_2_tmp)
if doc_file not in doc_file_to_sents:
doc_file_to_sents[doc_file] = [len(sents_all) - 1]
else:
doc_file_to_sents[doc_file] += [len(sents_all) - 1]
return sents_all,ners_all,ner_vocab,ner_1,ner_2,doc_file_to_sents
def data_transformation_doc(sents_list,ner_1_list,ner_2_list,ner_list,vocab_2_id,ner_2_id,word_unk_id,ner_to_id_1,ner_to_id_2):
"""
transform the raw data into numerics
"""
encode_res = []
for i,senti in enumerate(sents_list):
neri = ner_list[i]
ner_1_i = ner_1_list[i]
ner_2_i = ner_2_list[i]
ner_tmp = []
sent_tmp = []
ner_1_tmp = []
ner_2_tmp = []
for k, wordk in enumerate(senti):
nerk = neri[k]
try:
sent_tmp.append(vocab_2_id[wordk])
except:
sent_tmp.append(word_unk_id)
ner_tmp.append(ner_2_id[nerk])
ner_1_tmp.append(ner_to_id_1[ner_1_i[k]])
ner_2_tmp.append(ner_to_id_2[ner_2_i[k]])
encode_res.append([sent_tmp,ner_1_tmp,ner_2_tmp,ner_tmp])
return encode_res
def batch_generation_doc(doc_to_sents,enc_list,batch_size,max_doc_len,max_seq_len,vocab_2_id,ner_2_id, num_epoches=1):
# padding and trimming
ner_pad = ner_2_id['O']
word_pad = vocab_2_id['<PAD>']
valid_len_list = []
for i,linei in enumerate(enc_list):
senti = linei[0]
ner_1_i = linei[1]
ner_2_i = linei[2]
neri = linei[3]
valid_len_list.append(min(len(senti),max_seq_len))
senti = senti[:max_seq_len]
senti = senti + [word_pad] * max(0,max_seq_len-len(senti))
neri = neri[:max_seq_len]
neri = neri + [ner_pad] * max(0, max_seq_len - len(neri))
ner_1_i = ner_1_i[:max_seq_len]
ner_1_i = ner_1_i + [0] * max(0, max_seq_len - len(ner_1_i))
ner_2_i = ner_2_i[:max_seq_len]
ner_2_i = ner_2_i + [0] * max(0, max_seq_len - len(ner_2_i))
enc_list[i] = [senti,ner_1_i,ner_2_i,neri]
docs_all = []
for kk,dockk in enumerate(list(doc_to_sents.keys())):
sent_ids = doc_to_sents[dockk]
if len(sent_ids) <= max_doc_len:
sent_all = []
ner_1_all = []
ner_2_all = []
ner_all = []
valid_sents = len(sent_ids)
valid_words = []
for idi in sent_ids:
sent_all.append(enc_list[idi][0])
ner_1_all.append(enc_list[idi][1])
ner_2_all.append(enc_list[idi][2])
ner_all.append(enc_list[idi][3])
valid_words.append(valid_len_list[idi])
for kk in range(max_doc_len - valid_sents):
sent_all.append(enc_list[idi][0])
ner_1_all.append(enc_list[idi][1])
ner_2_all.append(enc_list[idi][2])
ner_all.append(enc_list[idi][3])
valid_words.append(valid_len_list[idi])
docs_all.append([sent_all,ner_1_all,ner_2_all,ner_all,valid_sents,valid_words])
else:
len_all = len(sent_ids)
ndocs_mini = int(np.ceil(len_all / max_doc_len))
for kk in range(ndocs_mini):
init_step = kk * max_doc_len
end_step = kk * max_doc_len + max_doc_len
ids_tmp = sent_ids[init_step:end_step]
sent_all = []
ner_1_all = []
ner_2_all = []
ner_all = []
valid_sents = len(ids_tmp)
valid_words = []
for idi in ids_tmp:
sent_all.append(enc_list[idi][0])
ner_1_all.append(enc_list[idi][1])
ner_2_all.append(enc_list[idi][2])
ner_all.append(enc_list[idi][3])
valid_words.append(valid_len_list[idi])
for kk in range(max_doc_len - valid_sents):
sent_all.append(enc_list[idi][0])
ner_1_all.append(enc_list[idi][1])
ner_2_all.append(enc_list[idi][2])
ner_all.append(enc_list[idi][3])
valid_words.append(valid_len_list[idi])
docs_all.append([sent_all, ner_1_all, ner_2_all, ner_all, valid_sents, valid_words])
random.shuffle(docs_all)
batches_all = []
sent_alls = []
ner_1_alls = []
ner_2_alls = []
ner_alls = []
valid_sentss = []
valid_wordss = []
docs_all = docs_all * num_epoches
for k,dock in enumerate(docs_all):
if k % batch_size == 0 and k > 0:
batches_all.append([sent_alls,ner_1_alls,ner_2_alls,ner_alls,batch_size,valid_sentss,valid_wordss])
sent_alls = []
ner_1_alls = []
ner_2_alls = []
ner_alls = []
valid_sentss = []
valid_wordss = []
sent_alls.append(dock[0])
ner_1_alls.append(dock[1])
ner_2_alls.append(dock[2])
ner_alls.append(dock[3])
valid_sentss.append(dock[4])
valid_wordss.append(dock[5])
else:
sent_alls.append(dock[0])
ner_1_alls.append(dock[1])
ner_2_alls.append(dock[2])
ner_alls.append(dock[3])
valid_sentss.append(dock[4])
valid_wordss.append(dock[5])
# paste the final
len_valid = len(sent_alls)
if len_valid == batch_size:
batches_all.append([sent_alls, ner_1_alls, ner_2_alls, ner_alls, len_valid, valid_sentss, valid_wordss])
else:
sent_alls += [sent_alls[-1]] * (batch_size - len_valid)
ner_1_alls += [ner_1_alls[-1]] * (batch_size - len_valid)
ner_2_alls += [ner_2_alls[-1]] * (batch_size - len_valid)
ner_alls += [ner_alls[-1]] * (batch_size - len_valid)
valid_sentss += [valid_sentss[-1]] * (batch_size - len_valid)
valid_wordss += [valid_wordss[-1]] * (batch_size - len_valid)
batches_all.append([sent_alls, ner_1_alls, ner_2_alls, ner_alls, len_valid, valid_sentss, valid_wordss])
return batches_all
if __name__ == "__main__":
pass