gabrielaltay
commited on
Commit
•
e5e4e82
1
Parent(s):
1b67cf1
upload hubscripts/pubhealth_hub.py to hub from bigbio repo
Browse files- pubhealth.py +209 -0
pubhealth.py
ADDED
@@ -0,0 +1,209 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# coding=utf-8
|
2 |
+
# Copyright 2020 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 |
+
A dataset of 11,832 claims for fact- checking, which are related a range of health topics
|
17 |
+
including biomedical subjects (e.g., infectious diseases, stem cell research), government healthcare policy
|
18 |
+
(e.g., abortion, mental health, women’s health), and other public health-related stories
|
19 |
+
"""
|
20 |
+
|
21 |
+
import csv
|
22 |
+
import os
|
23 |
+
from pathlib import Path
|
24 |
+
|
25 |
+
import datasets
|
26 |
+
|
27 |
+
from .bigbiohub import pairs_features
|
28 |
+
from .bigbiohub import BigBioConfig
|
29 |
+
from .bigbiohub import Tasks
|
30 |
+
|
31 |
+
logger = datasets.utils.logging.get_logger(__name__)
|
32 |
+
|
33 |
+
_LANGUAGES = ['English']
|
34 |
+
_PUBMED = False
|
35 |
+
_LOCAL = False
|
36 |
+
_CITATION = """\
|
37 |
+
@article{kotonya2020explainable,
|
38 |
+
title={Explainable automated fact-checking for public health claims},
|
39 |
+
author={Kotonya, Neema and Toni, Francesca},
|
40 |
+
journal={arXiv preprint arXiv:2010.09926},
|
41 |
+
year={2020}
|
42 |
+
}
|
43 |
+
"""
|
44 |
+
|
45 |
+
_DATASETNAME = "pubhealth"
|
46 |
+
_DISPLAYNAME = "PUBHEALTH"
|
47 |
+
|
48 |
+
_DESCRIPTION = """\
|
49 |
+
A dataset of 11,832 claims for fact- checking, which are related a range of health topics
|
50 |
+
including biomedical subjects (e.g., infectious diseases, stem cell research), government healthcare policy
|
51 |
+
(e.g., abortion, mental health, women’s health), and other public health-related stories
|
52 |
+
"""
|
53 |
+
|
54 |
+
_HOMEPAGE = "https://github.com/neemakot/Health-Fact-Checking/tree/master/data"
|
55 |
+
|
56 |
+
_LICENSE = 'MIT License'
|
57 |
+
|
58 |
+
_URLs = {
|
59 |
+
_DATASETNAME: "https://drive.google.com/uc?export=download&id=1eTtRs5cUlBP5dXsx-FTAlmXuB6JQi2qj"
|
60 |
+
}
|
61 |
+
|
62 |
+
_SUPPORTED_TASKS = [Tasks.TEXT_CLASSIFICATION]
|
63 |
+
_SOURCE_VERSION = "1.0.0"
|
64 |
+
_BIGBIO_VERSION = "1.0.0"
|
65 |
+
|
66 |
+
_CLASSES = ["true", "false", "unproven", "mixture"]
|
67 |
+
|
68 |
+
|
69 |
+
class PUBHEALTHDataset(datasets.GeneratorBasedBuilder):
|
70 |
+
"""Pubhealth text classification dataset"""
|
71 |
+
|
72 |
+
SOURCE_VERSION = datasets.Version(_SOURCE_VERSION)
|
73 |
+
BIGBIO_VERSION = datasets.Version(_BIGBIO_VERSION)
|
74 |
+
|
75 |
+
BUILDER_CONFIGS = [
|
76 |
+
BigBioConfig(
|
77 |
+
name="pubhealth_source",
|
78 |
+
version=SOURCE_VERSION,
|
79 |
+
description="PUBHEALTH source schema",
|
80 |
+
schema="source",
|
81 |
+
subset_id="pubhealth",
|
82 |
+
),
|
83 |
+
BigBioConfig(
|
84 |
+
name="pubhealth_bigbio_pairs",
|
85 |
+
version=BIGBIO_VERSION,
|
86 |
+
description="PUBHEALTH BigBio schema",
|
87 |
+
schema="bigbio_pairs",
|
88 |
+
subset_id="pubhealth",
|
89 |
+
),
|
90 |
+
]
|
91 |
+
|
92 |
+
DEFAULT_CONFIG_NAME = "pubhealth_source"
|
93 |
+
|
94 |
+
def _info(self):
|
95 |
+
|
96 |
+
if self.config.schema == "source":
|
97 |
+
features = datasets.Features(
|
98 |
+
{
|
99 |
+
"claim_id": datasets.Value("string"),
|
100 |
+
"claim": datasets.Value("string"),
|
101 |
+
"date_published": datasets.Value("string"),
|
102 |
+
"explanation": datasets.Value("string"),
|
103 |
+
"fact_checkers": datasets.Value("string"),
|
104 |
+
"main_text": datasets.Value("string"),
|
105 |
+
"sources": datasets.Value("string"),
|
106 |
+
"label": datasets.ClassLabel(names=_CLASSES),
|
107 |
+
"subjects": datasets.Value("string"),
|
108 |
+
}
|
109 |
+
)
|
110 |
+
|
111 |
+
# Using in entailment schema
|
112 |
+
elif self.config.schema == "bigbio_pairs":
|
113 |
+
features = pairs_features
|
114 |
+
|
115 |
+
return datasets.DatasetInfo(
|
116 |
+
description=_DESCRIPTION,
|
117 |
+
features=features,
|
118 |
+
homepage=_HOMEPAGE,
|
119 |
+
license=str(_LICENSE),
|
120 |
+
citation=_CITATION,
|
121 |
+
)
|
122 |
+
|
123 |
+
def _split_generators(self, dl_manager):
|
124 |
+
"""Returns SplitGenerators."""
|
125 |
+
urls = _URLs[_DATASETNAME]
|
126 |
+
data_dir = Path(dl_manager.download_and_extract(urls))
|
127 |
+
|
128 |
+
return [
|
129 |
+
datasets.SplitGenerator(
|
130 |
+
name=datasets.Split.TRAIN,
|
131 |
+
gen_kwargs={
|
132 |
+
"filepath": os.path.join(data_dir, "PUBHEALTH/train.tsv"),
|
133 |
+
"split": "train",
|
134 |
+
},
|
135 |
+
),
|
136 |
+
datasets.SplitGenerator(
|
137 |
+
name=datasets.Split.TEST,
|
138 |
+
gen_kwargs={
|
139 |
+
"filepath": os.path.join(data_dir, "PUBHEALTH/test.tsv"),
|
140 |
+
"split": "test",
|
141 |
+
},
|
142 |
+
),
|
143 |
+
datasets.SplitGenerator(
|
144 |
+
name=datasets.Split.VALIDATION,
|
145 |
+
gen_kwargs={
|
146 |
+
"filepath": os.path.join(data_dir, "PUBHEALTH/dev.tsv"),
|
147 |
+
"split": "validation",
|
148 |
+
},
|
149 |
+
),
|
150 |
+
]
|
151 |
+
|
152 |
+
def _generate_examples(self, filepath, split):
|
153 |
+
"""Yields examples as (key, example) tuples."""
|
154 |
+
|
155 |
+
with open(filepath, encoding="utf-8") as csv_file:
|
156 |
+
csv_reader = csv.reader(
|
157 |
+
csv_file,
|
158 |
+
quotechar='"',
|
159 |
+
delimiter="\t",
|
160 |
+
quoting=csv.QUOTE_NONE,
|
161 |
+
skipinitialspace=True,
|
162 |
+
)
|
163 |
+
next(csv_reader, None) # remove column headers
|
164 |
+
for id_, row in enumerate(csv_reader):
|
165 |
+
# train.tsv/dev.tsv only has 9 columns
|
166 |
+
# test.tsv has an additional column at the beginning
|
167 |
+
# Some entries are malformed, will log skipped lines
|
168 |
+
if len(row) < 9:
|
169 |
+
logger.info("Line %s is malformed", id_)
|
170 |
+
continue
|
171 |
+
(
|
172 |
+
claim_id,
|
173 |
+
claim,
|
174 |
+
date_published,
|
175 |
+
explanation,
|
176 |
+
fact_checkers,
|
177 |
+
main_text,
|
178 |
+
sources,
|
179 |
+
label,
|
180 |
+
subjects,
|
181 |
+
) = row[
|
182 |
+
-9:
|
183 |
+
] # only take last 9 columns to fix test.tsv disparity
|
184 |
+
|
185 |
+
if label not in _CLASSES:
|
186 |
+
logger.info("Line %s is missing label", id_)
|
187 |
+
continue
|
188 |
+
|
189 |
+
if self.config.schema == "source":
|
190 |
+
yield id_, {
|
191 |
+
"claim_id": claim_id,
|
192 |
+
"claim": claim,
|
193 |
+
"date_published": date_published,
|
194 |
+
"explanation": explanation,
|
195 |
+
"fact_checkers": fact_checkers,
|
196 |
+
"main_text": main_text,
|
197 |
+
"sources": sources,
|
198 |
+
"label": label,
|
199 |
+
"subjects": subjects,
|
200 |
+
}
|
201 |
+
|
202 |
+
elif self.config.schema == "bigbio_pairs":
|
203 |
+
yield id_, {
|
204 |
+
"id": id_, # uid is an unique identifier for every record that starts from 0
|
205 |
+
"document_id": claim_id,
|
206 |
+
"text_1": claim,
|
207 |
+
"text_2": explanation,
|
208 |
+
"label": label,
|
209 |
+
}
|