gabrielaltay
commited on
Commit
·
be90a7a
1
Parent(s):
b17e23e
upload hubscripts/n2c2_2008_hub.py to hub from bigbio repo
Browse files- n2c2_2008.py +424 -0
n2c2_2008.py
ADDED
@@ -0,0 +1,424 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# coding=utf-8
|
2 |
+
# Copyright 2022 The HuggingFace Datasets Authors and the current dataset script contributor.
|
3 |
+
#
|
4 |
+
# Licensed under the Apache License, Version 2.0 (the "License");
|
5 |
+
# you may not use this file except in compliance with the License.
|
6 |
+
# You may obtain a copy of the License at
|
7 |
+
#
|
8 |
+
# http://www.apache.org/licenses/LICENSE-2.0
|
9 |
+
#
|
10 |
+
# Unless required by applicable law or agreed to in writing, software
|
11 |
+
# distributed under the License is distributed on an "AS IS" BASIS,
|
12 |
+
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
13 |
+
# See the License for the specific language governing permissions and
|
14 |
+
# limitations under the License.
|
15 |
+
|
16 |
+
|
17 |
+
"""
|
18 |
+
A dataset loader for the n2c2 2008 obesity and comorbidities dataset.
|
19 |
+
|
20 |
+
https://portal.dbmi.hms.harvard.edu/projects/n2c2-nlp/
|
21 |
+
|
22 |
+
The dataset consists of eight xml files,
|
23 |
+
|
24 |
+
* obesity_patient_records_training.xml
|
25 |
+
* obesity_patient_records_training2.xml
|
26 |
+
* obesity_standoff_annotations_training.xml
|
27 |
+
* obesity_standoff_annotations_training_addendum.xml
|
28 |
+
* obesity_standoff_annotations_training_addendum2.xml
|
29 |
+
* obesity_standoff_annotations_training_addendum3.xml
|
30 |
+
* obesity_patient_records_test.xml
|
31 |
+
* obesity_standoff_annotations_test.xml
|
32 |
+
|
33 |
+
containing patient records as well as textual and intuitive annotations.
|
34 |
+
|
35 |
+
|
36 |
+
The files comprising this dataset must be on the users local machine
|
37 |
+
in a single directory that is passed to `datasets.load_datset` via
|
38 |
+
the `data_dir` kwarg. This loader script will read the xml files
|
39 |
+
directly. For example, if the following directory structure exists
|
40 |
+
on the users local machine,
|
41 |
+
|
42 |
+
|
43 |
+
n2c2_2008
|
44 |
+
├── obesity_patient_records_training.xml
|
45 |
+
├── obesity_patient_records_training2.xml
|
46 |
+
├── obesity_standoff_annotations_training.xml
|
47 |
+
├── obesity_standoff_annotations_training_addendum.xml
|
48 |
+
├── obesity_standoff_annotations_training_addendum2.xml
|
49 |
+
├── obesity_standoff_annotations_training_addendum3.xml
|
50 |
+
├── obesity_patient_records_test.xml
|
51 |
+
├── obesity_standoff_annotations_test.xml
|
52 |
+
|
53 |
+
|
54 |
+
Data Access
|
55 |
+
|
56 |
+
from https://www.i2b2.org/NLP/DataSets/Main.php
|
57 |
+
|
58 |
+
"As always, you must register AND submit a DUA for access. If you previously
|
59 |
+
accessed the data sets here on i2b2.org, you will need to set a new password
|
60 |
+
for your account on the Data Portal, but your original DUA will be retained."
|
61 |
+
|
62 |
+
|
63 |
+
"""
|
64 |
+
|
65 |
+
import os
|
66 |
+
import xml.etree.ElementTree as et
|
67 |
+
from pathlib import Path
|
68 |
+
from typing import Dict, List, Tuple
|
69 |
+
|
70 |
+
import datasets
|
71 |
+
|
72 |
+
from .bigbiohub import text_features
|
73 |
+
from .bigbiohub import BigBioConfig
|
74 |
+
from .bigbiohub import Tasks
|
75 |
+
|
76 |
+
_DATASETNAME = "n2c2_2008"
|
77 |
+
_DISPLAYNAME = "n2c2 2008 Obesity"
|
78 |
+
|
79 |
+
# https://academic.oup.com/jamia/article/16/4/561/766997
|
80 |
+
_LANGUAGES = ['English']
|
81 |
+
_PUBMED = True
|
82 |
+
_LOCAL = True
|
83 |
+
_CITATION = """\
|
84 |
+
@article{uzuner2009recognizing,
|
85 |
+
author = {
|
86 |
+
Uzuner, Ozlem
|
87 |
+
},
|
88 |
+
title = {Recognizing Obesity and Comorbidities in Sparse Data},
|
89 |
+
journal = {Journal of the American Medical Informatics Association},
|
90 |
+
volume = {16},
|
91 |
+
number = {4},
|
92 |
+
pages = {561-570},
|
93 |
+
year = {2009},
|
94 |
+
month = {07},
|
95 |
+
url = {https://doi.org/10.1197/jamia.M3115},
|
96 |
+
doi = {10.1197/jamia.M3115},
|
97 |
+
eprint = {https://academic.oup.com/jamia/article-pdf/16/4/561/2302602/16-4-561.pdf}
|
98 |
+
}
|
99 |
+
"""
|
100 |
+
|
101 |
+
_DESCRIPTION = """\
|
102 |
+
The data for the n2c2 2008 obesity challenge consisted of discharge summaries from
|
103 |
+
the Partners HealthCare Research Patient Data Repository. These data were chosen
|
104 |
+
from the discharge summaries of patients who were overweight or diabetic and had
|
105 |
+
been hospitalized for obesity or diabetes sometime since 12/1/04. De-identification
|
106 |
+
was performed semi-automatically. All private health information was replaced with
|
107 |
+
synthetic identifiers.
|
108 |
+
|
109 |
+
The data for the challenge were annotated by two obesity experts from the
|
110 |
+
Massachusetts General Hospital Weight Center. The experts were given a textual task,
|
111 |
+
which asked them to classify each disease (see list of diseases above) as Present,
|
112 |
+
Absent, Questionable, or Unmentioned based on explicitly documented information in
|
113 |
+
the discharge summaries, e.g., the statement “the patient is obese”. The experts were
|
114 |
+
also given an intuitive task, which asked them to classify each disease as Present,
|
115 |
+
Absent, or Questionable by applying their intuition and judgment to information in
|
116 |
+
the discharge summaries.
|
117 |
+
"""
|
118 |
+
|
119 |
+
_HOMEPAGE = "https://portal.dbmi.hms.harvard.edu/projects/n2c2-nlp/"
|
120 |
+
|
121 |
+
_LICENSE = 'Data User Agreement'
|
122 |
+
|
123 |
+
_SUPPORTED_TASKS = [Tasks.TEXT_CLASSIFICATION]
|
124 |
+
|
125 |
+
_CLASS_NAMES = ["present", "absent", "unmentioned", "questionable"]
|
126 |
+
_disease_names = [
|
127 |
+
"Obesity",
|
128 |
+
"Asthma",
|
129 |
+
"CAD",
|
130 |
+
"CHF",
|
131 |
+
"Depression",
|
132 |
+
"Diabetes",
|
133 |
+
"Gallstones",
|
134 |
+
"GERD",
|
135 |
+
"Gout",
|
136 |
+
"Hypercholesterolemia",
|
137 |
+
"Hypertension",
|
138 |
+
"Hypertriglyceridemia",
|
139 |
+
"OA",
|
140 |
+
"OSA",
|
141 |
+
"PVD",
|
142 |
+
"Venous Insufficiency",
|
143 |
+
]
|
144 |
+
|
145 |
+
_SOURCE_VERSION = "1.0.0"
|
146 |
+
_BIGBIO_VERSION = "1.0.0"
|
147 |
+
|
148 |
+
|
149 |
+
def _map_labels(doc, task):
|
150 |
+
"""
|
151 |
+
Map obesity and comorbidity labels.
|
152 |
+
:param doc: a document indexde by id
|
153 |
+
:param task: textual or intuitive annotation task
|
154 |
+
"""
|
155 |
+
lmap = {"Y": "present", "N": "absent", "U": "unmentioned", "Q": "questionable"}
|
156 |
+
|
157 |
+
def _map_label(doc, task, label_name):
|
158 |
+
if label_name in doc[task].keys():
|
159 |
+
return lmap[doc[task][label_name]]
|
160 |
+
else:
|
161 |
+
return None
|
162 |
+
|
163 |
+
if task in doc.keys():
|
164 |
+
return {
|
165 |
+
"Obesity": _map_label(doc, task, "Obesity"),
|
166 |
+
"Asthma": _map_label(doc, task, "Asthma"),
|
167 |
+
"CAD": _map_label(doc, task, "CAD"),
|
168 |
+
"CHF": _map_label(doc, task, "CHF"),
|
169 |
+
"Depression": _map_label(doc, task, "Depression"),
|
170 |
+
"Diabetes": _map_label(doc, task, "Diabetes"),
|
171 |
+
"Gallstones": _map_label(doc, task, "Gallstones"),
|
172 |
+
"GERD": _map_label(doc, task, "GERD"),
|
173 |
+
"Gout": _map_label(doc, task, "Gout"),
|
174 |
+
"Hypercholesterolemia": _map_label(doc, task, "Hypercholesterolemia"),
|
175 |
+
"Hypertension": _map_label(doc, task, "Hypertension"),
|
176 |
+
"Hypertriglyceridemia": _map_label(doc, task, "Hypertriglyceridemia"),
|
177 |
+
"OA": _map_label(doc, task, "OA"),
|
178 |
+
"OSA": _map_label(doc, task, "OSA"),
|
179 |
+
"PVD": _map_label(doc, task, "PVD"),
|
180 |
+
"Venous Insufficiency": _map_label(doc, task, "Venous Insufficiency"),
|
181 |
+
}
|
182 |
+
else:
|
183 |
+
return {task: None}
|
184 |
+
|
185 |
+
|
186 |
+
def _read_xml(partition, data_dir):
|
187 |
+
"""
|
188 |
+
Load the data split.
|
189 |
+
:param partition: train/test
|
190 |
+
:param data_dir: train and test data directory
|
191 |
+
"""
|
192 |
+
documents = {}
|
193 |
+
all_diseases = set()
|
194 |
+
notes = tuple()
|
195 |
+
if partition == "train":
|
196 |
+
with open(data_dir / "obesity_patient_records_training.xml") as t1, open(
|
197 |
+
data_dir / "obesity_patient_records_training2.xml"
|
198 |
+
) as t2:
|
199 |
+
notes1 = t1.read().strip()
|
200 |
+
notes2 = t2.read().strip()
|
201 |
+
notes = (notes1, notes2)
|
202 |
+
elif partition == "test":
|
203 |
+
with open(data_dir / "obesity_patient_records_test.xml") as t1:
|
204 |
+
notes1 = t1.read().strip()
|
205 |
+
notes = (notes1,)
|
206 |
+
|
207 |
+
for file in notes:
|
208 |
+
root = et.fromstring(file)
|
209 |
+
root = root.findall("./docs")[0]
|
210 |
+
for document in root.findall("./doc"):
|
211 |
+
assert document.attrib["id"] not in documents
|
212 |
+
documents[document.attrib["id"]] = {}
|
213 |
+
documents[document.attrib["id"]]["text"] = document.findall("./text")[
|
214 |
+
0
|
215 |
+
].text
|
216 |
+
|
217 |
+
annotation_files = tuple()
|
218 |
+
if partition == "train":
|
219 |
+
with open(data_dir / "obesity_standoff_annotations_training.xml") as t1, open(
|
220 |
+
data_dir / "obesity_standoff_annotations_training_addendum.xml"
|
221 |
+
) as t2, open(
|
222 |
+
data_dir / "obesity_standoff_annotations_training_addendum2.xml"
|
223 |
+
) as t3, open(
|
224 |
+
data_dir / "obesity_standoff_annotations_training_addendum3.xml"
|
225 |
+
) as t4:
|
226 |
+
train1 = t1.read().strip()
|
227 |
+
train2 = t2.read().strip()
|
228 |
+
train3 = t3.read().strip()
|
229 |
+
train4 = t4.read().strip()
|
230 |
+
annotation_files = (train1, train2, train3, train4)
|
231 |
+
elif partition == "test":
|
232 |
+
with open(data_dir / "obesity_standoff_annotations_test.xml") as t1:
|
233 |
+
test1 = t1.read().strip()
|
234 |
+
annotation_files = (test1,)
|
235 |
+
|
236 |
+
for file in annotation_files:
|
237 |
+
root = et.fromstring(file)
|
238 |
+
for diseases_annotation in root.findall("./diseases"):
|
239 |
+
|
240 |
+
annotation_source = diseases_annotation.attrib["source"]
|
241 |
+
assert isinstance(annotation_source, str)
|
242 |
+
for disease in diseases_annotation.findall("./disease"):
|
243 |
+
disease_name = disease.attrib["name"]
|
244 |
+
all_diseases.add(disease_name)
|
245 |
+
for annotation in disease.findall("./doc"):
|
246 |
+
doc_id = annotation.attrib["id"]
|
247 |
+
if not annotation_source in documents[doc_id]:
|
248 |
+
documents[doc_id][annotation_source] = {}
|
249 |
+
assert doc_id in documents
|
250 |
+
judgment = annotation.attrib["judgment"]
|
251 |
+
documents[doc_id][annotation_source][disease_name] = judgment
|
252 |
+
return [
|
253 |
+
{
|
254 |
+
"document_id": str(id),
|
255 |
+
"text": documents[id]["text"],
|
256 |
+
"textual": _map_labels(documents[id], "textual"),
|
257 |
+
"intuitive": _map_labels(documents[id], "intuitive"),
|
258 |
+
}
|
259 |
+
for id in documents
|
260 |
+
]
|
261 |
+
|
262 |
+
|
263 |
+
class N2C22008ObesityDataset(datasets.GeneratorBasedBuilder):
|
264 |
+
"""n2c2 2008 obesity and comorbidities recognition task"""
|
265 |
+
|
266 |
+
SOURCE_VERSION = datasets.Version(_SOURCE_VERSION)
|
267 |
+
BIGBIO_VERSION = datasets.Version(_BIGBIO_VERSION)
|
268 |
+
|
269 |
+
BUILDER_CONFIGS = [
|
270 |
+
BigBioConfig(
|
271 |
+
name="n2c2_2008_source",
|
272 |
+
version=SOURCE_VERSION,
|
273 |
+
description="n2c2_2008 source schema",
|
274 |
+
schema="source",
|
275 |
+
subset_id="n2c2_2008",
|
276 |
+
),
|
277 |
+
BigBioConfig(
|
278 |
+
name="n2c2_2008_bigbio_text",
|
279 |
+
version=BIGBIO_VERSION,
|
280 |
+
description="n2c2_2008 BigBio schema",
|
281 |
+
schema="bigbio_text",
|
282 |
+
subset_id="n2c2_2008",
|
283 |
+
),
|
284 |
+
]
|
285 |
+
|
286 |
+
DEFAULT_CONFIG_NAME = "n2c2_2008_source"
|
287 |
+
|
288 |
+
def _info(self) -> datasets.DatasetInfo:
|
289 |
+
|
290 |
+
if self.config.schema == "source":
|
291 |
+
features = datasets.Features(
|
292 |
+
{
|
293 |
+
"document_id": datasets.Value("string"),
|
294 |
+
"text": datasets.Value("string"),
|
295 |
+
"labels": [
|
296 |
+
{
|
297 |
+
"annotation": datasets.ClassLabel(
|
298 |
+
names=["textual", "intuitive"]
|
299 |
+
),
|
300 |
+
"disease_name": datasets.ClassLabel(names=_disease_names),
|
301 |
+
"label": datasets.ClassLabel(names=_CLASS_NAMES),
|
302 |
+
}
|
303 |
+
],
|
304 |
+
}
|
305 |
+
)
|
306 |
+
|
307 |
+
elif self.config.schema == "bigbio_text":
|
308 |
+
features = text_features
|
309 |
+
|
310 |
+
return datasets.DatasetInfo(
|
311 |
+
description=_DESCRIPTION,
|
312 |
+
features=features,
|
313 |
+
homepage=_HOMEPAGE,
|
314 |
+
license=str(_LICENSE),
|
315 |
+
citation=_CITATION,
|
316 |
+
)
|
317 |
+
|
318 |
+
def _split_generators(
|
319 |
+
self, dl_manager: datasets.DownloadManager
|
320 |
+
) -> List[datasets.SplitGenerator]:
|
321 |
+
"""Returns SplitGenerators."""
|
322 |
+
|
323 |
+
if self.config.data_dir is None:
|
324 |
+
raise ValueError(
|
325 |
+
"This is a local dataset. Please pass the data_dir kwarg to load_dataset."
|
326 |
+
)
|
327 |
+
else:
|
328 |
+
data_dir = self.config.data_dir
|
329 |
+
|
330 |
+
return [
|
331 |
+
datasets.SplitGenerator(
|
332 |
+
name=datasets.Split.TRAIN,
|
333 |
+
gen_kwargs={
|
334 |
+
"data_dir": data_dir,
|
335 |
+
"split": "train",
|
336 |
+
},
|
337 |
+
),
|
338 |
+
datasets.SplitGenerator(
|
339 |
+
name=datasets.Split.TEST,
|
340 |
+
gen_kwargs={
|
341 |
+
"data_dir": data_dir,
|
342 |
+
"split": "test",
|
343 |
+
},
|
344 |
+
),
|
345 |
+
]
|
346 |
+
|
347 |
+
@staticmethod
|
348 |
+
def _get_source_sample(sample):
|
349 |
+
textual_labels = [
|
350 |
+
("textual", disease_name, sample["textual"][disease_name])
|
351 |
+
for disease_name in sample["textual"].keys()
|
352 |
+
if sample["textual"][disease_name]
|
353 |
+
]
|
354 |
+
intuitive_labels = [
|
355 |
+
("intuitive", disease_name, sample["intuitive"][disease_name])
|
356 |
+
for disease_name in sample["intuitive"].keys()
|
357 |
+
if sample["intuitive"][disease_name]
|
358 |
+
]
|
359 |
+
|
360 |
+
return {
|
361 |
+
"document_id": sample["document_id"],
|
362 |
+
"text": sample["text"],
|
363 |
+
"labels": [
|
364 |
+
{
|
365 |
+
"annotation": label[0],
|
366 |
+
"disease_name": label[1],
|
367 |
+
"label": label[2],
|
368 |
+
}
|
369 |
+
for label in textual_labels + intuitive_labels
|
370 |
+
],
|
371 |
+
}
|
372 |
+
|
373 |
+
@staticmethod
|
374 |
+
def _get_bigbio_sample(sample_id, sample):
|
375 |
+
textual_labels = [
|
376 |
+
("textual", disease_name, sample["textual"][disease_name])
|
377 |
+
for disease_name in sample["textual"].keys()
|
378 |
+
if sample["textual"][disease_name]
|
379 |
+
]
|
380 |
+
intuitive_labels = [
|
381 |
+
("intuitive", disease_name, sample["intuitive"][disease_name])
|
382 |
+
for disease_name in sample["intuitive"].keys()
|
383 |
+
if sample["intuitive"][disease_name]
|
384 |
+
]
|
385 |
+
|
386 |
+
return {
|
387 |
+
"id": str(sample_id),
|
388 |
+
"document_id": sample["document_id"],
|
389 |
+
"text": sample["text"],
|
390 |
+
"labels": [
|
391 |
+
{
|
392 |
+
"annotation": label[0],
|
393 |
+
"disease_name": label[1],
|
394 |
+
"label": label[2],
|
395 |
+
}
|
396 |
+
for label in textual_labels + intuitive_labels
|
397 |
+
],
|
398 |
+
}
|
399 |
+
|
400 |
+
def _generate_examples(self, data_dir, split: str) -> Tuple[int, Dict]:
|
401 |
+
"""Yields examples as (key, example) tuples."""
|
402 |
+
|
403 |
+
data_dir = Path(data_dir).resolve()
|
404 |
+
if split == "train":
|
405 |
+
_id = 0
|
406 |
+
samples = _read_xml(split, data_dir)
|
407 |
+
for sample in samples:
|
408 |
+
if self.config.schema == "source":
|
409 |
+
yield _id, self._get_source_sample(sample)
|
410 |
+
|
411 |
+
elif self.config.schema == "bigbio_text":
|
412 |
+
yield _id, self._get_bigbio_sample(_id, sample)
|
413 |
+
_id += 1
|
414 |
+
|
415 |
+
elif split == "test":
|
416 |
+
_id = 0
|
417 |
+
samples = _read_xml(split, data_dir)
|
418 |
+
for sample in samples:
|
419 |
+
if self.config.schema == "source":
|
420 |
+
yield _id, self._get_source_sample(sample)
|
421 |
+
|
422 |
+
elif self.config.schema == "bigbio_text":
|
423 |
+
yield _id, self._get_bigbio_sample(_id, sample)
|
424 |
+
_id += 1
|