keep pos tags
Browse files- wikiner.py +287 -5
wikiner.py
CHANGED
@@ -62,6 +62,285 @@ class Wikiner(datasets.GeneratorBasedBuilder):
|
|
62 |
{
|
63 |
"id": datasets.Value("string"),
|
64 |
"tokens": datasets.Sequence(datasets.Value("string")),
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
65 |
"ner_tags": datasets.Sequence(
|
66 |
datasets.features.ClassLabel(
|
67 |
names=[
|
@@ -100,29 +379,32 @@ class Wikiner(datasets.GeneratorBasedBuilder):
|
|
100 |
|
101 |
def _generate_examples(self, filepath):
|
102 |
logging.info("⏳ Generating examples from = %s", filepath)
|
103 |
-
with open(filepath, encoding="utf-8") as f:
|
104 |
guid = 0
|
105 |
-
tokens = []
|
106 |
-
ner_tags = []
|
107 |
for line in f:
|
108 |
if line.startswith("-DOCSTART-") or line == "" or line == "\n":
|
109 |
if tokens:
|
110 |
yield guid, {
|
111 |
"id": str(guid),
|
112 |
"tokens": tokens,
|
|
|
113 |
"ner_tags": ner_tags,
|
114 |
}
|
115 |
guid += 1
|
116 |
tokens = []
|
|
|
117 |
ner_tags = []
|
118 |
else:
|
119 |
-
splits = line.split(
|
120 |
tokens.append(splits[0])
|
121 |
-
|
|
|
122 |
# last example
|
123 |
if tokens:
|
124 |
yield guid, {
|
125 |
"id": str(guid),
|
126 |
"tokens": tokens,
|
|
|
127 |
"ner_tags": ner_tags,
|
128 |
}
|
|
|
62 |
{
|
63 |
"id": datasets.Value("string"),
|
64 |
"tokens": datasets.Sequence(datasets.Value("string")),
|
65 |
+
"pos_tags": datasets.Sequence(
|
66 |
+
datasets.features.ClassLabel(
|
67 |
+
names=[
|
68 |
+
"ACRNM",
|
69 |
+
"ADJ",
|
70 |
+
"ADV",
|
71 |
+
"ALFS",
|
72 |
+
"ART",
|
73 |
+
"BACKSLASH",
|
74 |
+
"CARD",
|
75 |
+
"CC",
|
76 |
+
"CCAD",
|
77 |
+
"CCNEG",
|
78 |
+
"CM",
|
79 |
+
"CODE",
|
80 |
+
"COLON",
|
81 |
+
"CQUE",
|
82 |
+
"CSUBF",
|
83 |
+
"CSUBI",
|
84 |
+
"CSUBX",
|
85 |
+
"DM",
|
86 |
+
"DOTS",
|
87 |
+
"FS",
|
88 |
+
"INT",
|
89 |
+
"LP",
|
90 |
+
"NC",
|
91 |
+
"NEG",
|
92 |
+
"NMEA",
|
93 |
+
"NMON",
|
94 |
+
"NP",
|
95 |
+
"ORD",
|
96 |
+
"PAL",
|
97 |
+
"PDEL",
|
98 |
+
"PE",
|
99 |
+
"PERCT",
|
100 |
+
"PPC",
|
101 |
+
"PPO",
|
102 |
+
"PPX",
|
103 |
+
"PREP",
|
104 |
+
"QT",
|
105 |
+
"QU",
|
106 |
+
"REL",
|
107 |
+
"RP",
|
108 |
+
"SE",
|
109 |
+
"SEMICOLON",
|
110 |
+
"SLASH",
|
111 |
+
"SYM",
|
112 |
+
"UMMX",
|
113 |
+
"VCLIfin",
|
114 |
+
"VCLIger",
|
115 |
+
"VCLIinf",
|
116 |
+
"VEadj",
|
117 |
+
"VEfin",
|
118 |
+
"VEger",
|
119 |
+
"VEinf",
|
120 |
+
"VHadj",
|
121 |
+
"VHfin",
|
122 |
+
"VHger",
|
123 |
+
"VHinf",
|
124 |
+
"VLadj",
|
125 |
+
"VLfin",
|
126 |
+
"VLger",
|
127 |
+
"VLinf",
|
128 |
+
"VMadj",
|
129 |
+
"VMfin",
|
130 |
+
"VMger",
|
131 |
+
"VMinf",
|
132 |
+
"VSadj",
|
133 |
+
"VSfin",
|
134 |
+
"VSger",
|
135 |
+
"VSinf",
|
136 |
+
]
|
137 |
+
)
|
138 |
+
),
|
139 |
+
"ner_tags": datasets.Sequence(
|
140 |
+
datasets.features.ClassLabel(
|
141 |
+
names=[
|
142 |
+
"O",
|
143 |
+
"B-PER",
|
144 |
+
"I-PER",
|
145 |
+
"B-ORG",
|
146 |
+
"I-ORG",
|
147 |
+
"B-LOC",
|
148 |
+
"I-LOC",
|
149 |
+
"B-MISC",
|
150 |
+
"I-MISC",
|
151 |
+
]
|
152 |
+
)
|
153 |
+
),
|
154 |
+
}
|
155 |
+
),
|
156 |
+
supervised_keys=None,
|
157 |
+
homepage=_URL,
|
158 |
+
citation=_CITATION,
|
159 |
+
)
|
160 |
+
|
161 |
+
def _split_generators(self, dl_manager):
|
162 |
+
"""Returns SplitGenerators."""
|
163 |
+
urls_to_download = {
|
164 |
+
"train": f"{_URL}{_TRAINING_FILE}",
|
165 |
+
}
|
166 |
+
downloaded_files = dl_manager.download_and_extract(urls_to_download)
|
167 |
+
|
168 |
+
return [
|
169 |
+
datasets.SplitGenerator(
|
170 |
+
name=datasets.Split.TRAIN,
|
171 |
+
gen_kwargs={"filepath": downloaded_files["train"]},
|
172 |
+
),
|
173 |
+
]
|
174 |
+
|
175 |
+
def _generate_examples(self, filepath):
|
176 |
+
logging.info("⏳ Generating examples from = %s", filepath)
|
177 |
+
with open(filepath, encoding="utf-8") as f:
|
178 |
+
guid = 0
|
179 |
+
tokens, pos_tags, ner_tags = [], [], []
|
180 |
+
for line in f:
|
181 |
+
if line.startswith("-DOCSTART-") or line == "" or line == "\n":
|
182 |
+
if tokens:
|
183 |
+
yield guid, {
|
184 |
+
"id": str(guid),
|
185 |
+
"tokens": tokens,
|
186 |
+
"pos_tags": pos_tags,
|
187 |
+
"ner_tags": ner_tags,
|
188 |
+
}
|
189 |
+
guid += 1
|
190 |
+
tokens = []
|
191 |
+
pos_tags = []
|
192 |
+
ner_tags = []
|
193 |
+
else:
|
194 |
+
splits = line.split(" ")
|
195 |
+
tokens.append(splits[0])
|
196 |
+
pos_tags.append(splits[1])
|
197 |
+
ner_tags.append(splits[2].rstrip())
|
198 |
+
# last example
|
199 |
+
if tokens:
|
200 |
+
yield guid, {
|
201 |
+
"id": str(guid),
|
202 |
+
"tokens": tokens,
|
203 |
+
"pos_tags": pos_tags,
|
204 |
+
"ner_tags": ner_tags,
|
205 |
+
}
|
206 |
+
# coding=utf-8
|
207 |
+
# Copyright 2020 HuggingFace Datasets Authors.
|
208 |
+
#
|
209 |
+
# Licensed under the Apache License, Version 2.0 (the "License");
|
210 |
+
# you may not use this file except in compliance with the License.
|
211 |
+
# You may obtain a copy of the License at
|
212 |
+
#
|
213 |
+
# http://www.apache.org/licenses/LICENSE-2.0
|
214 |
+
#
|
215 |
+
# Unless required by applicable law or agreed to in writing, software
|
216 |
+
# distributed under the License is distributed on an "AS IS" BASIS,
|
217 |
+
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
218 |
+
# See the License for the specific language governing permissions and
|
219 |
+
# limitations under the License.
|
220 |
+
|
221 |
+
import logging
|
222 |
+
|
223 |
+
import datasets
|
224 |
+
|
225 |
+
|
226 |
+
_CITATION = """\
|
227 |
+
@inproceedings{,
|
228 |
+
title = "",
|
229 |
+
author = "Garagiola, Nazareno",
|
230 |
+
year = "2022",
|
231 |
+
url = ""
|
232 |
+
}
|
233 |
+
"""
|
234 |
+
|
235 |
+
_DESCRIPTION = """Dataset used to train a NER model"""
|
236 |
+
|
237 |
+
_URL = "https://raw.githubusercontent.com/NazaGara/betoNER/main/data/wikiner/"
|
238 |
+
_TRAINING_FILE = "train.conllu"
|
239 |
+
|
240 |
+
|
241 |
+
class WikinerConfig(datasets.BuilderConfig):
|
242 |
+
"""BuilderConfig"""
|
243 |
+
|
244 |
+
def __init__(self, **kwargs):
|
245 |
+
"""BuilderConfig
|
246 |
+
Args:
|
247 |
+
**kwargs: keyword arguments forwarded to super.
|
248 |
+
"""
|
249 |
+
super(WikinerConfig, self).__init__(**kwargs)
|
250 |
+
|
251 |
+
|
252 |
+
class Wikiner(datasets.GeneratorBasedBuilder):
|
253 |
+
"""Wikiner dataset."""
|
254 |
+
|
255 |
+
BUILDER_CONFIGS = [
|
256 |
+
WikinerConfig(
|
257 |
+
name="wikiner",
|
258 |
+
version=datasets.Version("1.0.0"),
|
259 |
+
description="wikiner dataset",
|
260 |
+
),
|
261 |
+
]
|
262 |
+
|
263 |
+
def _info(self):
|
264 |
+
return datasets.DatasetInfo(
|
265 |
+
description=_DESCRIPTION,
|
266 |
+
features=datasets.Features(
|
267 |
+
{
|
268 |
+
"id": datasets.Value("string"),
|
269 |
+
"tokens": datasets.Sequence(datasets.Value("string")),
|
270 |
+
"pos_tags": datasets.Sequence(
|
271 |
+
datasets.features.ClassLabel(
|
272 |
+
names=[
|
273 |
+
"ACRNM",
|
274 |
+
"ADJ",
|
275 |
+
"ADV",
|
276 |
+
"ALFS",
|
277 |
+
"ART",
|
278 |
+
"BACKSLASH",
|
279 |
+
"CARD",
|
280 |
+
"CC",
|
281 |
+
"CCAD",
|
282 |
+
"CCNEG",
|
283 |
+
"CM",
|
284 |
+
"CODE",
|
285 |
+
"COLON",
|
286 |
+
"CQUE",
|
287 |
+
"CSUBF",
|
288 |
+
"CSUBI",
|
289 |
+
"CSUBX",
|
290 |
+
"DM",
|
291 |
+
"DOTS",
|
292 |
+
"FS",
|
293 |
+
"INT",
|
294 |
+
"LP",
|
295 |
+
"NC",
|
296 |
+
"NEG",
|
297 |
+
"NMEA",
|
298 |
+
"NMON",
|
299 |
+
"NP",
|
300 |
+
"ORD",
|
301 |
+
"PAL",
|
302 |
+
"PDEL",
|
303 |
+
"PE",
|
304 |
+
"PERCT",
|
305 |
+
"PPC",
|
306 |
+
"PPO",
|
307 |
+
"PPX",
|
308 |
+
"PREP",
|
309 |
+
"QT",
|
310 |
+
"QU",
|
311 |
+
"REL",
|
312 |
+
"RP",
|
313 |
+
"SE",
|
314 |
+
"SEMICOLON",
|
315 |
+
"SLASH",
|
316 |
+
"SYM",
|
317 |
+
"UMMX",
|
318 |
+
"VCLIfin",
|
319 |
+
"VCLIger",
|
320 |
+
"VCLIinf",
|
321 |
+
"VEadj",
|
322 |
+
"VEfin",
|
323 |
+
"VEger",
|
324 |
+
"VEinf",
|
325 |
+
"VHadj",
|
326 |
+
"VHfin",
|
327 |
+
"VHger",
|
328 |
+
"VHinf",
|
329 |
+
"VLadj",
|
330 |
+
"VLfin",
|
331 |
+
"VLger",
|
332 |
+
"VLinf",
|
333 |
+
"VMadj",
|
334 |
+
"VMfin",
|
335 |
+
"VMger",
|
336 |
+
"VMinf",
|
337 |
+
"VSadj",
|
338 |
+
"VSfin",
|
339 |
+
"VSger",
|
340 |
+
"VSinf",
|
341 |
+
]
|
342 |
+
)
|
343 |
+
),
|
344 |
"ner_tags": datasets.Sequence(
|
345 |
datasets.features.ClassLabel(
|
346 |
names=[
|
|
|
379 |
|
380 |
def _generate_examples(self, filepath):
|
381 |
logging.info("⏳ Generating examples from = %s", filepath)
|
382 |
+
with open(filepath, encoding="utf-8") as f:
|
383 |
guid = 0
|
384 |
+
tokens, pos_tags, ner_tags = [], [], []
|
|
|
385 |
for line in f:
|
386 |
if line.startswith("-DOCSTART-") or line == "" or line == "\n":
|
387 |
if tokens:
|
388 |
yield guid, {
|
389 |
"id": str(guid),
|
390 |
"tokens": tokens,
|
391 |
+
"pos_tags": pos_tags,
|
392 |
"ner_tags": ner_tags,
|
393 |
}
|
394 |
guid += 1
|
395 |
tokens = []
|
396 |
+
pos_tags = []
|
397 |
ner_tags = []
|
398 |
else:
|
399 |
+
splits = line.split(" ")
|
400 |
tokens.append(splits[0])
|
401 |
+
pos_tags.append(splits[1])
|
402 |
+
ner_tags.append(splits[2].rstrip())
|
403 |
# last example
|
404 |
if tokens:
|
405 |
yield guid, {
|
406 |
"id": str(guid),
|
407 |
"tokens": tokens,
|
408 |
+
"pos_tags": pos_tags,
|
409 |
"ner_tags": ner_tags,
|
410 |
}
|