forked from bigscience-workshop/biomedical
-
Notifications
You must be signed in to change notification settings - Fork 0
/
paramed.py
197 lines (169 loc) · 6.86 KB
/
paramed.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
# coding=utf-8
# Copyright 2020 The HuggingFace Datasets Authors and the current dataset script contributor.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
"""
NEJM is a Chinese-English parallel corpus crawled from the New England Journal of Medicine website.
English articles are distributed through https://www.nejm.org/ and Chinese articles are distributed through
http://nejmqianyan.cn/. The corpus contains all article pairs (around 2000 pairs) since 2011.
The script loads dataset in bigbio schema (using schemas/text-to-text) AND/OR source (default) schema
"""
import os # useful for paths
from typing import Dict, Iterable, List
import datasets
from .bigbiohub import text2text_features
from .bigbiohub import BigBioConfig
from .bigbiohub import Tasks
logger = datasets.logging.get_logger(__name__)
_LANGUAGES = ['English', 'Chinese']
_PUBMED = False
_LOCAL = False
_CITATION = """\
@article{liu2021paramed,
author = {Liu, Boxiang and Huang, Liang},
title = {ParaMed: a parallel corpus for English–Chinese translation in the biomedical domain},
journal = {BMC Medical Informatics and Decision Making},
volume = {21},
year = {2021},
url = {https://bmcmedinformdecismak.biomedcentral.com/articles/10.1186/s12911-021-01621-8},
doi = {10.1186/s12911-021-01621-8}
}
"""
_DATASETNAME = "paramed"
_DISPLAYNAME = "ParaMed"
_DESCRIPTION = """\
NEJM is a Chinese-English parallel corpus crawled from the New England Journal of Medicine website.
English articles are distributed through https://www.nejm.org/ and Chinese articles are distributed through
http://nejmqianyan.cn/. The corpus contains all article pairs (around 2000 pairs) since 2011.
"""
_HOMEPAGE = "https://github.com/boxiangliu/ParaMed"
_LICENSE = 'Creative Commons Attribution 4.0 International'
_URLs = {
"source": "https://github.com/boxiangliu/ParaMed/blob/master/data/nejm-open-access.tar.gz?raw=true",
"bigbio_t2t": "https://github.com/boxiangliu/ParaMed/blob/master/data/nejm-open-access.tar.gz?raw=true",
}
_SUPPORTED_TASKS = [Tasks.TRANSLATION]
_SOURCE_VERSION = "1.0.0"
_BIGBIO_VERSION = "1.0.0"
_DATA_DIR = "./processed_data/open_access/open_access"
class ParamedDataset(datasets.GeneratorBasedBuilder):
"""Write a short docstring documenting what this dataset is"""
SOURCE_VERSION = datasets.Version(_SOURCE_VERSION)
BIGBIO_VERSION = datasets.Version(_BIGBIO_VERSION)
BUILDER_CONFIGS = [
BigBioConfig(
name="paramed_source",
version=SOURCE_VERSION,
description="Paramed source schema",
schema="source",
subset_id="paramed",
),
BigBioConfig(
name="paramed_bigbio_t2t",
version=BIGBIO_VERSION,
description="Paramed BigBio schema",
schema="bigbio_t2t",
subset_id="paramed",
),
]
DEFAULT_CONFIG_NAME = "paramed_source"
def _info(self):
if self.config.schema == "source":
features = datasets.Features(
{
"document_id": datasets.Value("string"),
"text_1": datasets.Value("string"),
"text_2": datasets.Value("string"),
"text_1_name": datasets.Value("string"),
"text_2_name": datasets.Value("string"),
}
)
elif self.config.schema == "bigbio_t2t":
features = text2text_features
return datasets.DatasetInfo(
# This is the description that will appear on the datasets page.
description=_DESCRIPTION,
features=features,
homepage=_HOMEPAGE,
license=str(_LICENSE),
citation=_CITATION,
)
def _split_generators(
self, dl_manager: datasets.DownloadManager
) -> List[datasets.SplitGenerator]:
my_urls = _URLs[self.config.schema]
data_dir = os.path.join(dl_manager.download_and_extract(my_urls), _DATA_DIR)
print(data_dir)
return [
datasets.SplitGenerator(
name=datasets.Split.TRAIN,
gen_kwargs={
"filepath": data_dir,
"zh_file": os.path.join(data_dir, "nejm.train.zh"),
"en_file": os.path.join(data_dir, "nejm.train.en"),
"split": "train",
},
),
datasets.SplitGenerator(
name=datasets.Split.VALIDATION,
gen_kwargs={
"filepath": data_dir,
"zh_file": os.path.join(data_dir, "nejm.dev.zh"),
"en_file": os.path.join(data_dir, "nejm.dev.en"),
"split": "dev",
},
),
datasets.SplitGenerator(
name=datasets.Split.TEST,
gen_kwargs={
"filepath": data_dir,
"zh_file": os.path.join(data_dir, "nejm.test.zh"),
"en_file": os.path.join(data_dir, "nejm.test.en"),
"split": "test",
},
),
]
def _generate_examples(self, filepath, zh_file, en_file, split):
logger.info("generating examples from = %s", filepath)
zh_file = open(zh_file, "r")
en_file = open(en_file, "r")
zh_file.seek(0)
en_file.seek(0)
zh_lines = zh_file.readlines()
en_lines = en_file.readlines()
assert len(en_lines) == len(zh_lines), "Line mismatch"
if self.config.schema == "source":
for key, (zh_line, en_line) in enumerate(zip(zh_lines, en_lines)):
yield key, {
"document_id": str(key),
"text_1": zh_line,
"text_2": en_line,
"text_1_name": "zh",
"text_2_name": "en",
}
zh_file.close()
en_file.close()
elif self.config.schema == "bigbio_t2t":
uid = 0
for key, (zh_line, en_line) in enumerate(zip(zh_lines, en_lines)):
uid += 1
yield key, {
"id": str(uid),
"document_id": str(key),
"text_1": zh_line,
"text_2": en_line,
"text_1_name": "zh",
"text_2_name": "en",
}
zh_file.close()
en_file.close()