-
Notifications
You must be signed in to change notification settings - Fork 0
/
data.py
116 lines (90 loc) · 3.33 KB
/
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
from datasets import Dataset, ClassLabel, Sequence, Features, Value, DatasetDict
from collections import defaultdict
import glob
with open('./data/First_Phase_Release/answer.txt', 'r', encoding='utf-8-sig') as f:
answers = f.read().splitlines()
with open('./data/Second_Phase_Dataset/answer.txt', 'r', encoding='utf-8-sig') as f:
answers += f.read().splitlines()
with open('./data/Validation_Dataset_Answer/answer.txt', 'r', encoding='utf-8-sig') as f:
answers += f.read().splitlines()
data_dict = {}
label_dict = defaultdict(list)
class_label = ['O']
filter_label = [
'DOCTOR',
'DATE',
'IDNUM',
'HOSPITAL',
'MEDICALRECORD',
'PATIENT',
'TIME',
'DEPARTMENT',
'CITY',
'ZIP',
'STREET',
'STATE',
'AGE',
'ORGANIZATION'
]
for answer in answers:
line = answer.split("\t")
if line[1] in filter_label:
label_dict[line[0]].append([line[1], line[4]])
if f"B-{line[1]}" not in class_label:
class_label.append(f"B-{line[1]}")
class_label.append(f"I-{line[1]}")
def create_ner_dataset(file_content, label_data):
labels = {}
for label_pair in label_data:
label_type, label_value = label_pair
labels[label_type] = label_value
lines = file_content.split('\n')
file_token = []
file_annotation = []
for line in lines:
line = line.strip()
if line:
tokens = line.split()
annotations = []
for token in tokens:
annotation = class_label.index('O')
for label_type, label_value in labels.items():
if label_value in token:
annotation = class_label.index(f"B-{label_type}")
break
annotations.append(annotation)
file_token.extend(tokens)
file_annotation.extend(annotations)
if len(file_annotation) != len(file_token):
assert False, "Length of annotation and token is not equal"
return file_token, file_annotation
def create_dataset(paths):
file_list = []
for path in paths:
file_list += glob.glob(path + '*')
ids = []
tokens_list = []
ner_tags_list = []
for file in file_list:
idx = file.split('/')[-1].split('.')[0]
with open(file, 'r', encoding='utf-8-sig') as f:
data = f.read()
tokens, ner_tags = create_ner_dataset(data, label_dict[idx])
ids.append(idx)
tokens_list.append(tokens)
ner_tags_list.append(ner_tags)
data_dict = {
'tokens': tokens_list,
'ner_tags': ner_tags_list,
}
tags = ClassLabel(num_classes=len(class_label), names=class_label)
dataset_structure = {"ner_tags":Sequence(tags),
'tokens': Sequence(feature=Value(dtype='string'))}
return Dataset.from_dict(mapping=data_dict, features=Features(dataset_structure))
def get_train_dataset():
train_folder_path = ["./data/First_Phase_Release/First_Phase_Text_Dataset/", "./data/Second_Phase_Dataset/Second_Phase_Text_Dataset/"]
val_folder_path = ["./data/First_Phase_Release/Validation_Release/"]
train_dataset = create_dataset(train_folder_path)
val_dataset = create_dataset(val_folder_path)
raw_datasets = DatasetDict({'train': train_dataset, 'validation': val_dataset})
return raw_datasets