-
Notifications
You must be signed in to change notification settings - Fork 13
/
preprocess.py
198 lines (174 loc) · 6.89 KB
/
preprocess.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
#!/usr/bin/python
# -*- coding: utf-8 -*-
import csv
import re
import distance
import sys
import pandas as pd
"""
该文件为预处理文件，主要进行以下几个预处理：
1.清除无用的信息
2.如果待预测实体不在文本的前512中，将预测实体所在的文本提前到前512中
3.将文本中出现的实体，添加上“<”，“>”，来突出实体
4.将含有多条实体的数据切分成多条只预测一个实体的数据
5.截断文本（取前512）
"""
max_seq_length = 512
def clean_space(text):
""""
处理多余的空格
"""
match_regex = re.compile(u'[\u4e00-\u9fa5。\.,，:：《》、\(\)（）]{1} +(?<![a-zA-Z])|\d+ +| +\d+|[a-z A-Z]+')
should_replace_list = match_regex.findall(text)
order_replace_list = sorted(should_replace_list,key=lambda i:len(i),reverse=True)
for i in order_replace_list:
if i == u' ':
continue
new_i = i.strip()
text = text.replace(i,new_i)
return text
def clean(text):
"""
清除无用的信息
:param text:
:return:
"""
if type(text) != str:
return text
text = clean_space(text)
text = re.sub(r'\n', '', text)
text = re.sub(r'\{IMG:[0-9]+\}', '', text)
text = re.sub(r'\?{2,}', '', text)
text = re.sub(r'[0-9a-zA-Z]{100,}', '', text)
text = re.sub(r'<.*?>', '', text)
text = re.sub(r'http(s?):[/a-zA-Z0-9.=&?_#]+', '', text)
text = re.sub(r'“', '', text)
text = re.sub(r'”', '', text)
text = re.sub(r'—{5,}', '', text)
text = re.sub(r'？{2,}', '', text)
text = re.sub(r'●', '', text)
text = re.sub(r'【图】', '', text)
text = re.sub(r'[0-9]+[-|.|/|年][0-9]{2}[-|.|/|月][0-9]{2}日?', '', text)
text = re.sub(r' ', '', text)
text = re.sub(r'[0-9]{15,}', '', text)
text = re.sub(r'"', '', text)
return text
def match(title, text):
if type(title) == float or type(text) == float:
return 1
strs = list(distance.lcsubstrings(title, text))
if len(strs) > 0:
return len(strs[0]) / len(title)
else:
return 0
def get_entity_sentence(entity, text):
"""
找出预测实体所在的文本
:param entity:
:param text:
:return:
"""
index = text.find(entity)
if index > 512:
split_symbol = ['.', '。', '!', '！', '?', '？', '；', ';', ' ', '\t', '\n']
for i in range(50):
if text[index - 20 - i - 1] in split_symbol:
return ''.join(text[index - 20 - i: len(text)])
else:
return None
def process(filename, data_path, mode='train'):
"""
数据预处理主函数
:param filename:
:param mode:
:return:
"""
header = []
rows = []
with open(filename, 'r', encoding='utf-8-sig') as f:
f_csv = csv.reader(f)
for i, row in enumerate(f_csv):
if i == 0:
if mode == 'train':
header = [row[0], row[2], 'clean_entity'] + row[3:5]
else:
header = [row[0], row[2], 'clean_entity', row[3]]
else:
text_id, title, text = row[0:3]
if row[3] != '':
entities = row[3].split(';')
else:
entities = ['']
if mode == 'train':
if row[5] != '':
key_entities = clean(row[5]).split(';')
else:
key_entities = []
if len(text) == 0 or type(text) == float:
text = title
text = clean(text)
for index, entity in enumerate(entities):
clean_entity = clean(entity)
entity_in_text = get_entity_sentence(clean_entity, text)
new_text = text
if entity_in_text != None:
new_text = entity_in_text
if clean_entity not in text:
new_text = clean_entity + '。' + text
if entity == '' or entity != entity or type(entity) == float:
continue
if clean_entity == '' or clean_entity != clean_entity or type(clean_entity) == float:
continue
new_text = new_text.replace(clean_entity, '<' + clean_entity + '>')
if mode == 'train':
if clean_entity in key_entities:
negative = 1
else:
negative = 0
if mode == 'train':
new_entities = entities.copy()
new_entities[index] = '<' + clean_entity + '>'
new_entities_sub = []
for i in range(index-10, index + 10):
if i >= 0 and i < len(new_entities):
new_entities_sub.append(new_entities[i])
new_entity = ';'.join(new_entities_sub)
rows.append([text_id, new_text, new_entity, entity, negative])
else:
new_entities = entities.copy()
new_entities[index] = '<' + clean_entity + '>'
new_entities_sub = []
for i in range(index-10, index + 10):
if i >= 0 and i < len(new_entities):
new_entities_sub.append(new_entities[i])
new_entity = ';'.join(new_entities_sub)
rows.append([text_id, new_text, new_entity, entity])
with open(data_path, 'w', encoding='utf-8-sig', newline='') as f:
f_csv = csv.writer(f)
f_csv.writerow(header)
for row in rows:
f_csv.writerow(row)
def merge_round_1_round_2():
"""
合并初赛和复赛的数据
:return:
"""
train_round_1 = pd.read_csv('./datasets/Train_Data.csv')
train_round_2 = pd.read_csv('./datasets/Round2_train.csv')
merge_data = pd.concat([train_round_2, train_round_1], ignore_index=True)
merge_data = merge_data.drop_duplicates(keep='first', subset=['text'])
merge_data.to_csv('./datasets/round_1_2_train_data.csv', encoding='utf-8', index=False)
if __name__ == '__main__':
merge_round_1_round_2()
if len(sys.argv) < 2:
# 训练集的预处理
process('./datasets/round_1_2_train_data.csv', './datasets/preprocess_round_1_2_train_data.csv', 'train')
# 测试集的预处理
process('./datasets/round2_test.csv', './datasets/preprocess_round2_test.csv', 'test')
else:
train_path = sys.argv[1]
test_path = sys.argv[2]
# 训练集的预处理
process(train_path, './datasets/preprocess_round_1_2_train_data.csv', 'train')
# 测试集的预处理
process(test_path, './datasets/preprocess_round2_test.csv', 'test')