Datasets:
Tasks:
Text Classification
Sub-tasks:
multi-label-classification
Languages:
English
Size:
10K<n<100K
License:
Peixian Wang
commited on
Commit
·
1826f48
1
Parent(s):
de6b5ec
add rtGender loader
Browse files- rtGender.py +272 -0
rtGender.py
ADDED
@@ -0,0 +1,272 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
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 |
+
"""Class for loading datafrom rtGender"""
|
16 |
+
|
17 |
+
from __future__ import absolute_import, division, print_function
|
18 |
+
|
19 |
+
import csv
|
20 |
+
from enum import Enum
|
21 |
+
import os
|
22 |
+
|
23 |
+
import datasets
|
24 |
+
|
25 |
+
|
26 |
+
_CITATION = """\
|
27 |
+
@inproceedings{voigt-etal-2018-rtgender,
|
28 |
+
title = "{R}t{G}ender: A Corpus for Studying Differential Responses to Gender",
|
29 |
+
author = "Voigt, Rob and
|
30 |
+
Jurgens, David and
|
31 |
+
Prabhakaran, Vinodkumar and
|
32 |
+
Jurafsky, Dan and
|
33 |
+
Tsvetkov, Yulia",
|
34 |
+
booktitle = "Proceedings of the Eleventh International Conference on Language Resources and Evaluation ({LREC} 2018)",
|
35 |
+
month = may,
|
36 |
+
year = "2018",
|
37 |
+
address = "Miyazaki, Japan",
|
38 |
+
publisher = "European Language Resources Association (ELRA)",
|
39 |
+
url = "https://www.aclweb.org/anthology/L18-1445",
|
40 |
+
}
|
41 |
+
"""
|
42 |
+
|
43 |
+
_DESCRIPTION = """\
|
44 |
+
RtGender is a corpus for studying responses to gender online, including posts and responses from Facebook, TED, Fitocracy, and Reddit where the gender of the source poster/speaker is known.
|
45 |
+
"""
|
46 |
+
|
47 |
+
_HOMEPAGE = "https://nlp.stanford.edu/robvoigt/rtgender/#contact"
|
48 |
+
|
49 |
+
_LICENSE = "Research Only"
|
50 |
+
|
51 |
+
_URL = "https://nlp.stanford.edu/robvoigt/rtgender/rtgender.tar.gz"
|
52 |
+
|
53 |
+
|
54 |
+
class Config(Enum):
|
55 |
+
ANNOTATIONS = "annotations"
|
56 |
+
POSTS = "posts"
|
57 |
+
RESPONSES = "responses"
|
58 |
+
FB_POLI = "fb_politicians"
|
59 |
+
FB_PUB = "fb_public"
|
60 |
+
TED = "ted"
|
61 |
+
FITOCRACY = "fitocracy"
|
62 |
+
REDDIT = "reddit"
|
63 |
+
|
64 |
+
|
65 |
+
class rtGender(datasets.GeneratorBasedBuilder):
|
66 |
+
"""TODO: Short description of my dataset."""
|
67 |
+
|
68 |
+
VERSION = datasets.Version("1.1.0")
|
69 |
+
|
70 |
+
# This is an example of a dataset with multiple configurations.
|
71 |
+
# If you don't want/need to define several sub-sets in your dataset,
|
72 |
+
# just remove the BUILDER_CONFIG_CLASS and the BUILDER_CONFIGS attributes.
|
73 |
+
|
74 |
+
# If you need to make complex sub-parts in the datasets with configurable options
|
75 |
+
# You can create your own builder configuration class to store attribute, inheriting from datasets.BuilderConfig
|
76 |
+
# BUILDER_CONFIG_CLASS = MyBuilderConfig
|
77 |
+
|
78 |
+
# You will be able to load one or the other configurations in the following list with
|
79 |
+
# data = datasets.load_dataset('my_dataset', 'first_domain')
|
80 |
+
# data = datasets.load_dataset('my_dataset', 'second_domain')
|
81 |
+
BUILDER_CONFIGS = [
|
82 |
+
datasets.BuilderConfig(
|
83 |
+
name=str(Config.ANNOTATIONS),
|
84 |
+
version=VERSION,
|
85 |
+
description="Covers 30k annotations",
|
86 |
+
),
|
87 |
+
datasets.BuilderConfig(
|
88 |
+
name=str(Config.POSTS),
|
89 |
+
version=VERSION,
|
90 |
+
description="This part of my dataset covers a second domain",
|
91 |
+
),
|
92 |
+
datasets.BuilderConfig(
|
93 |
+
name=str(Config.RESPONSES),
|
94 |
+
version=VERSION,
|
95 |
+
description="This part of my dataset covers a second domain",
|
96 |
+
)
|
97 |
+
]
|
98 |
+
DEFAULT_CONFIG_NAME = str(Config.ANNOTATIONS) # It's not mandatory to have a default configuration. Just use one if it make sense.
|
99 |
+
|
100 |
+
POSTS_FEATURES = {
|
101 |
+
"source": datasets.Value("string"),
|
102 |
+
"op_id": datasets.Value("string"),
|
103 |
+
"op_gender": datasets.Value("string"),
|
104 |
+
"post_id": datasets.Value("string"),
|
105 |
+
"post_text": datasets.Value("string"),
|
106 |
+
"post_type": datasets.Value("string"), # only for fb
|
107 |
+
"subreddit": datasets.Value("string"), # only for reddit
|
108 |
+
"op_gender_visible": datasets.Value("string"), # only for reddit
|
109 |
+
}
|
110 |
+
|
111 |
+
RESPONSES_FEATURES = {
|
112 |
+
"source": datasets.Value("string"),
|
113 |
+
"op_id": datasets.Value("string"),
|
114 |
+
"op_gender": datasets.Value("string"),
|
115 |
+
"post_id": datasets.Value("string"),
|
116 |
+
"responder_id": datasets.Value("string"),
|
117 |
+
"response_text": datasets.Value("string"),
|
118 |
+
"op_name": datasets.Value("string"), # only for fb
|
119 |
+
"op_category": datasets.Value("string"), # only for fb
|
120 |
+
"responder_gender": datasets.Value("string"), # only for fitocracy and reddit
|
121 |
+
"responder_gender_visible": datasets.Value("string"), # only for reddit
|
122 |
+
"subreddit": datasets.Value("string"),
|
123 |
+
}
|
124 |
+
|
125 |
+
ANNOTATION_FEATURES = {
|
126 |
+
"source": datasets.Value("string"),
|
127 |
+
"op_gender": datasets.Value("string"),
|
128 |
+
"post_text": datasets.Value("string"),
|
129 |
+
"response_text": datasets.Value("string"),
|
130 |
+
"sentiment": datasets.Value("string"),
|
131 |
+
"relevance": datasets.Value("string"),
|
132 |
+
}
|
133 |
+
|
134 |
+
def _info(self):
|
135 |
+
|
136 |
+
if (
|
137 |
+
self.config.name == Config.ANNOTATIONS
|
138 |
+
): # This is the name of the configuration selected in BUILDER_CONFIGS above
|
139 |
+
features = datasets.Features(self.ANNOTATION_FEATURES)
|
140 |
+
elif self.config.name == Config.POSTS:
|
141 |
+
features = datasets.Features(self.POSTS_FEATURES)
|
142 |
+
else:
|
143 |
+
features = datasets.Features(self.RESPONSES_FEATURES)
|
144 |
+
return datasets.DatasetInfo(
|
145 |
+
# This is the description that will appear on the datasets page.
|
146 |
+
description=_DESCRIPTION,
|
147 |
+
# This defines the different columns of the dataset and their types
|
148 |
+
features=features, # Here we define them above because they are different between the two configurations
|
149 |
+
# If there's a common (input, target) tuple from the features,
|
150 |
+
# specify them here. They'll be used if as_supervised=True in
|
151 |
+
# builder.as_dataset.
|
152 |
+
supervised_keys=None,
|
153 |
+
# Homepage of the dataset for documentation
|
154 |
+
homepage=_HOMEPAGE,
|
155 |
+
# License for the dataset if available
|
156 |
+
license=_LICENSE,
|
157 |
+
# Citation for the dataset
|
158 |
+
citation=_CITATION,
|
159 |
+
)
|
160 |
+
|
161 |
+
def _split_generators(self, dl_manager):
|
162 |
+
"""Returns SplitGenerators."""
|
163 |
+
# If several configurations are possible (listed in BUILDER_CONFIGS), the configuration selected by the user is in self.config.name
|
164 |
+
|
165 |
+
# dl_manager is a datasets.download.DownloadManager that can be used to download and extract URLs
|
166 |
+
# It can accept any type or nested list/dict and will give back the same structure with the url replaced with path to local files.
|
167 |
+
# By default the archives will be extracted and a path to a cached folder where they are extracted is returned instead of the archive
|
168 |
+
data_dir = dl_manager.download_and_extract(_URL)
|
169 |
+
if self.config.name == Config.ANNOTATIONS:
|
170 |
+
files = ["annotations.csv"]
|
171 |
+
elif self.config.name == Config.POSTS:
|
172 |
+
files = [
|
173 |
+
"facebook_congress_posts.csv",
|
174 |
+
"facebook_wiki_posts.csv",
|
175 |
+
"fitocracy_posts.csv",
|
176 |
+
"reddit_posts.csv",
|
177 |
+
]
|
178 |
+
else:
|
179 |
+
files = [
|
180 |
+
"facebook_congress_responses.csv",
|
181 |
+
"facebook_wiki_responses.csv",
|
182 |
+
"fitocracy_responses.csv",
|
183 |
+
"reddit_responses.csv",
|
184 |
+
"ted_responses.csv",
|
185 |
+
]
|
186 |
+
return [
|
187 |
+
datasets.SplitGenerator(
|
188 |
+
name=datasets.Split.TRAIN,
|
189 |
+
# These kwargs will be passed to _generate_examples
|
190 |
+
gen_kwargs={
|
191 |
+
"filepaths": files,
|
192 |
+
"split": "train",
|
193 |
+
},
|
194 |
+
),
|
195 |
+
]
|
196 |
+
|
197 |
+
def _generate_examples(
|
198 |
+
self,
|
199 |
+
filepaths,
|
200 |
+
split, # method parameters are unpacked from `gen_kwargs` as given in `_split_generators`
|
201 |
+
):
|
202 |
+
""" Yields examples as (key, example) tuples. """
|
203 |
+
# This method handles input defined in _split_generators to yield (key, example) tuples from the dataset.
|
204 |
+
# The `key` is here for legacy reason (tfds) and is not important in itself.
|
205 |
+
files = []
|
206 |
+
readers = {}
|
207 |
+
for fp in filepaths:
|
208 |
+
f = open(fp, encoding="utf-8")
|
209 |
+
reader = csv.reader(f)
|
210 |
+
next(reader)
|
211 |
+
readers[fp.replace(".csv", "")] = reader
|
212 |
+
files.append(f)
|
213 |
+
|
214 |
+
id_ = 0
|
215 |
+
for reader_name, reader in readers.items():
|
216 |
+
for row in reader:
|
217 |
+
if self.config.name == Config.ANNOTATIONS:
|
218 |
+
yield id_, {
|
219 |
+
"source": row[0],
|
220 |
+
"op_gender": row[1],
|
221 |
+
"post_text": row[2],
|
222 |
+
"response_text": row[3],
|
223 |
+
"sentiment": row[4],
|
224 |
+
"relevance": row[5],
|
225 |
+
}
|
226 |
+
elif self.config.name == Config.POSTS:
|
227 |
+
r = {
|
228 |
+
"source": reader_name,
|
229 |
+
"op_id": row[0],
|
230 |
+
"op_gender": row[1],
|
231 |
+
"post_id": row[2],
|
232 |
+
"post_text": row[3],
|
233 |
+
"post_type": None,
|
234 |
+
"subreddit": None,
|
235 |
+
"op_gender_visible": None,
|
236 |
+
}
|
237 |
+
if "facebook" in reader_name:
|
238 |
+
r["post_type"] = row[4]
|
239 |
+
elif "reddit" in reader_name:
|
240 |
+
r["subreddit"] = row[4]
|
241 |
+
r["op_gender_visible"] = row[5]
|
242 |
+
|
243 |
+
yield id_, r
|
244 |
+
|
245 |
+
else:
|
246 |
+
r = {
|
247 |
+
"source": reader_name,
|
248 |
+
"op_id": row[0],
|
249 |
+
"op_gender": row[1],
|
250 |
+
"post_id": row[2],
|
251 |
+
"responder_id": row[3],
|
252 |
+
"response_text": row[4],
|
253 |
+
"op_name": None,
|
254 |
+
"op_category": None,
|
255 |
+
"responder_gender": None,
|
256 |
+
"responder_gender_visible": None,
|
257 |
+
"subreddit": None
|
258 |
+
}
|
259 |
+
if "facebook" in reader_name:
|
260 |
+
r["op_name"] = row[5]
|
261 |
+
r["op_category"] = row[6]
|
262 |
+
elif "fitocracy" in reader_name:
|
263 |
+
r["responder_gender"] = row[5]
|
264 |
+
elif "reddit" in reader_name:
|
265 |
+
r["subreddit"] = row[5]
|
266 |
+
r["responder_gender"] = row[6]
|
267 |
+
r["responder_gender_visible"] = row[7]
|
268 |
+
yield id_, r
|
269 |
+
id_ += 1
|
270 |
+
|
271 |
+
for fd in files:
|
272 |
+
fd.close()
|