-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathprepare_finetune_data.py
206 lines (171 loc) · 7.17 KB
/
prepare_finetune_data.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
import pandas as pd
from processors.utils import InputExample
from model import tokenization_albert
from tqdm import tqdm
# id2label = {0: '财经', 1: '房产', 2: '家居', 3: '教育', 4: '科技', 5: '时尚', 6: '时政', 7: '游戏', 8: '娱乐', 9: '体育'}
id2risk = {0: '高风险', 1: '中风险', 2: '可公开', 3: '低风险', 4: '中风险', 5: '低风险', 6: '高风险', 7: '低风险', 8: '可公开', 9: '可公开'}
label2risk = {'财经': '高风险', '房产': '中风险', '家居': '可公开', '教育': '低风险', '科技': '中风险', '时尚': '低风险', '时政': '高风险', '游戏': '低风险',
'娱乐': '可公开', '体育': '可公开'}
label2id = {'财经': 0, '房产': 1, '家居': 2, '教育': 3, '科技': 4, '时尚': 5, '时政': 6, '游戏': 7, '娱乐': 8, '体育': 9}
def is_chinese(uchar):
"""判断一个unicode是否是汉字"""
if uchar >= u'\u4e00' and uchar <= u'\u9fa5':
return True
else:
return False
def is_number(uchar):
"""判断一个unicode是否是数字"""
if uchar >= u'\u0030' and uchar <= u'\u0039':
return True
else:
return False
def is_alphabet(uchar):
"""判断一个unicode是否是英文字母"""
if (uchar >= u'\u0041' and uchar <= u'\u005a') or (uchar >= u'\u0061' and uchar <= u'\u007a'):
return True
else:
return False
def is_other(uchar):
"""判断是否非汉字，数字和英文字符"""
if not (is_chinese(uchar) or is_number(uchar) or is_alphabet(uchar)):
return True
else:
return False
def get_all_stopword():
words_set = set()
# path_list = ['baidu', 'cn', 'hit', 'scu', 'news']
path_list = ['baidu', 'cn', 'hit', 'scu']
for i in path_list:
path = './stopword/' + i + '_stopwords.txt'
with open(path, 'r', encoding='utf-8') as f:
data = f.readlines()
for _word in data:
words_set.add(_word.strip())
return words_set
def get_labeled_data_from_csv(path):
inputexample_list = []
df = pd.read_csv(path, encoding='utf-8')
for index, row in df.iterrows():
inputexample_list.append(
InputExample(guid=row["id"], text_a=row["content"].strip().replace(' ', ''), label=row["class_label"]))
return inputexample_list
def get_unlabeled_data_from_csv(path):
inputexample_list = []
df = pd.read_csv(path, encoding='utf-8')
for index, row in df.iterrows():
inputexample_list.append(
InputExample(guid=row["id"], text_a=row["content"].strip().replace(' ', '')))
return inputexample_list
def get_train_examples(flag='original'):
if flag == 'original':
path = './finetunedata/labeled_data.csv'
elif flag == 'extract':
path = 'finetunedata/labeled_data_extract_clean.csv'
else:
path = ''
assert path != ''
# path = './finetunedata/train_data.csv'
print('train data path', path)
return get_labeled_data_from_csv(path)
def get_dev_examples(flag='original'):
if flag == 'original':
path = './finetunedata/unlabeled_data_clean.csv'
elif flag == 'extract':
path = 'finetunedata/unlabeled_data_labeled_extract_clean.csv'
else:
path = ''
assert path != ''
print('dev data path', path)
return get_unlabeled_data_from_csv(path)
def get_test_examples(flag='original'):
if flag == 'original':
path = './finetunedata/test_data_clean.csv'
elif flag == 'extract':
path = 'finetunedata/test_data_extract_clean.csv'
else:
path = ''
assert path != ''
print('test data path', path)
return get_unlabeled_data_from_csv(path)
def get_all_example(flag='extract'):
train_example = get_train_examples(flag)
dev_example = get_dev_examples(flag)
test_example = get_test_examples(flag)
all_example = []
all_example.extend(train_example)
all_example.extend(dev_example)
all_example.extend(test_example)
return all_example
def get_unk_token():
# 补全字典
all_example = get_all_example()
unk_dict = {}
vocab_file = './prev_trained_model/vocab.txt'
tokenizer = tokenization_albert.FullTokenizer(vocab_file=vocab_file, do_lower_case=True, spm_model_file=None)
for example in tqdm(all_example):
for word in tokenizer.tokenize(example.text_a):
if word in unk_dict:
unk_dict[word] += 1
else:
unk_dict[word] = 1
print(len(unk_dict))
dict_sort = sorted(unk_dict.items(), key=lambda x: x[1], reverse=True)
# print(dict_sort)
with open('./prev_trained_model/unk_vocab.txt', 'w', encoding='utf-8') as f:
for i in dict_sort:
f.write(str(i) + '\n')
def get_unclean_token():
# 获取纯英文、纯中文、纯数字之外的字符
other_dict = {}
all_example = get_all_example()
tokenizer = tokenization_albert.BasicTokenizer(do_lower_case=True)
for example in tqdm(all_example):
tokens = tokenizer.tokenize(example.text_a)
for token in tokens:
if is_chinese(token) or str.isdigit(token) or str.isalpha(token):
pass
else:
if token in other_dict:
other_dict[token] += 1
else:
other_dict[token] = 1
print(len(other_dict))
dict_sort = sorted(other_dict.items(), key=lambda x: x[1], reverse=True)
# print(dict_sort)
with open('./prev_trained_model/unclean_token.txt', 'w', encoding='utf-8') as f:
for i in dict_sort:
f.write(str(i) + '\n')
def to_standard_train_data():
path = './finetunedata/labeled_data_clean.csv'
cont_list = []
df = pd.read_csv(path, encoding='utf-8')
for index, row in df.iterrows():
# print(row["class_label"])
cont_list.append(
{'id': row["id"], 'class_label': row["class_label"], 'content': row["content"].strip().replace(' ', '')})
df_s = pd.DataFrame(cont_list, columns=['id', 'class_label', 'content'])
df_s.to_csv('./finetunedata/labeled_data_clean_1.csv', columns=['id', 'class_label', 'content'], index=False,
header=True,
encoding='utf-8')
print("end")
print(df_s.columns)
def load_train_dev_data():
path_train = './finetunedata/labeled_data_clean.csv'
path_dev = './finetunedata/valid_uns_new.csv'
# path_train = './finetunedata/train_uns_fair.csv'
# path_dev = './finetunedata/valid_uns_fair.csv'
# path_train = './finetunedata/train_dataset5.csv'
# path_dev = './finetunedata/valid_dataset5.csv'
inputexample_train = get_labeled_data_from_csv(path_train)
inputexample_valid = get_labeled_data_from_csv(path_dev)
# print(len(inputexample_train))
# print(len(inputexample_valid))
return inputexample_train, inputexample_valid
if __name__ == '__main__':
# print(get_train_examples())
# print(get_test_examples())
# get_dev_examples()
# get_unk_token()
# get_unclean_token()
# to_standard_train_data()
load_train_dev_data()