Datasets:
Tasks:
Token Classification
Sub-tasks:
named-entity-recognition
Languages:
English
Size:
10K<n<100K
License:
Delete loading script
Browse files- conllpp.py +0 -228
conllpp.py
DELETED
@@ -1,228 +0,0 @@
|
|
1 |
-
# coding=utf-8
|
2 |
-
# Copyright 2020 HuggingFace Datasets Authors.
|
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 |
-
# Lint as: python3
|
17 |
-
"""CrossWeigh: Training Named Entity Tagger from Imperfect Annotations"""
|
18 |
-
|
19 |
-
import logging
|
20 |
-
|
21 |
-
import datasets
|
22 |
-
|
23 |
-
|
24 |
-
_CITATION = """\
|
25 |
-
@inproceedings{wang2019crossweigh,
|
26 |
-
title={CrossWeigh: Training Named Entity Tagger from Imperfect Annotations},
|
27 |
-
author={Wang, Zihan and Shang, Jingbo and Liu, Liyuan and Lu, Lihao and Liu, Jiacheng and Han, Jiawei},
|
28 |
-
booktitle={Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP)},
|
29 |
-
pages={5157--5166},
|
30 |
-
year={2019}
|
31 |
-
}
|
32 |
-
"""
|
33 |
-
|
34 |
-
_DESCRIPTION = """\
|
35 |
-
CoNLLpp is a corrected version of the CoNLL2003 NER dataset where labels of 5.38% of the sentences in the test set
|
36 |
-
have been manually corrected. The training set and development set are included for completeness.
|
37 |
-
For more details see https://www.aclweb.org/anthology/D19-1519/ and https://github.com/ZihanWangKi/CrossWeigh
|
38 |
-
"""
|
39 |
-
|
40 |
-
_URL = "https://github.com/ZihanWangKi/CrossWeigh/raw/master/data/"
|
41 |
-
_TRAINING_FILE = "conllpp_train.txt"
|
42 |
-
_DEV_FILE = "conllpp_dev.txt"
|
43 |
-
_TEST_FILE = "conllpp_test.txt"
|
44 |
-
|
45 |
-
|
46 |
-
class ConllppConfig(datasets.BuilderConfig):
|
47 |
-
"""BuilderConfig for Conll2003"""
|
48 |
-
|
49 |
-
def __init__(self, **kwargs):
|
50 |
-
"""BuilderConfig forConll2003.
|
51 |
-
Args:
|
52 |
-
**kwargs: keyword arguments forwarded to super.
|
53 |
-
"""
|
54 |
-
super(ConllppConfig, self).__init__(**kwargs)
|
55 |
-
|
56 |
-
|
57 |
-
class Conllpp(datasets.GeneratorBasedBuilder):
|
58 |
-
"""Conllpp dataset."""
|
59 |
-
|
60 |
-
BUILDER_CONFIGS = [
|
61 |
-
ConllppConfig(name="conllpp", version=datasets.Version("1.0.0"), description="Conllpp dataset"),
|
62 |
-
]
|
63 |
-
|
64 |
-
def _info(self):
|
65 |
-
return datasets.DatasetInfo(
|
66 |
-
description=_DESCRIPTION,
|
67 |
-
features=datasets.Features(
|
68 |
-
{
|
69 |
-
"id": datasets.Value("string"),
|
70 |
-
"tokens": datasets.Sequence(datasets.Value("string")),
|
71 |
-
"pos_tags": datasets.Sequence(
|
72 |
-
datasets.features.ClassLabel(
|
73 |
-
names=[
|
74 |
-
'"',
|
75 |
-
"''",
|
76 |
-
"#",
|
77 |
-
"$",
|
78 |
-
"(",
|
79 |
-
")",
|
80 |
-
",",
|
81 |
-
".",
|
82 |
-
":",
|
83 |
-
"``",
|
84 |
-
"CC",
|
85 |
-
"CD",
|
86 |
-
"DT",
|
87 |
-
"EX",
|
88 |
-
"FW",
|
89 |
-
"IN",
|
90 |
-
"JJ",
|
91 |
-
"JJR",
|
92 |
-
"JJS",
|
93 |
-
"LS",
|
94 |
-
"MD",
|
95 |
-
"NN",
|
96 |
-
"NNP",
|
97 |
-
"NNPS",
|
98 |
-
"NNS",
|
99 |
-
"NN|SYM",
|
100 |
-
"PDT",
|
101 |
-
"POS",
|
102 |
-
"PRP",
|
103 |
-
"PRP$",
|
104 |
-
"RB",
|
105 |
-
"RBR",
|
106 |
-
"RBS",
|
107 |
-
"RP",
|
108 |
-
"SYM",
|
109 |
-
"TO",
|
110 |
-
"UH",
|
111 |
-
"VB",
|
112 |
-
"VBD",
|
113 |
-
"VBG",
|
114 |
-
"VBN",
|
115 |
-
"VBP",
|
116 |
-
"VBZ",
|
117 |
-
"WDT",
|
118 |
-
"WP",
|
119 |
-
"WP$",
|
120 |
-
"WRB",
|
121 |
-
]
|
122 |
-
)
|
123 |
-
),
|
124 |
-
"chunk_tags": datasets.Sequence(
|
125 |
-
datasets.features.ClassLabel(
|
126 |
-
names=[
|
127 |
-
"O",
|
128 |
-
"B-ADJP",
|
129 |
-
"I-ADJP",
|
130 |
-
"B-ADVP",
|
131 |
-
"I-ADVP",
|
132 |
-
"B-CONJP",
|
133 |
-
"I-CONJP",
|
134 |
-
"B-INTJ",
|
135 |
-
"I-INTJ",
|
136 |
-
"B-LST",
|
137 |
-
"I-LST",
|
138 |
-
"B-NP",
|
139 |
-
"I-NP",
|
140 |
-
"B-PP",
|
141 |
-
"I-PP",
|
142 |
-
"B-PRT",
|
143 |
-
"I-PRT",
|
144 |
-
"B-SBAR",
|
145 |
-
"I-SBAR",
|
146 |
-
"B-UCP",
|
147 |
-
"I-UCP",
|
148 |
-
"B-VP",
|
149 |
-
"I-VP",
|
150 |
-
]
|
151 |
-
)
|
152 |
-
),
|
153 |
-
"ner_tags": datasets.Sequence(
|
154 |
-
datasets.features.ClassLabel(
|
155 |
-
names=[
|
156 |
-
"O",
|
157 |
-
"B-PER",
|
158 |
-
"I-PER",
|
159 |
-
"B-ORG",
|
160 |
-
"I-ORG",
|
161 |
-
"B-LOC",
|
162 |
-
"I-LOC",
|
163 |
-
"B-MISC",
|
164 |
-
"I-MISC",
|
165 |
-
]
|
166 |
-
)
|
167 |
-
),
|
168 |
-
}
|
169 |
-
),
|
170 |
-
supervised_keys=None,
|
171 |
-
homepage="https://github.com/ZihanWangKi/CrossWeigh",
|
172 |
-
citation=_CITATION,
|
173 |
-
)
|
174 |
-
|
175 |
-
def _split_generators(self, dl_manager):
|
176 |
-
"""Returns SplitGenerators."""
|
177 |
-
urls_to_download = {
|
178 |
-
"train": f"{_URL}{_TRAINING_FILE}",
|
179 |
-
"dev": f"{_URL}{_DEV_FILE}",
|
180 |
-
"test": f"{_URL}{_TEST_FILE}",
|
181 |
-
}
|
182 |
-
downloaded_files = dl_manager.download_and_extract(urls_to_download)
|
183 |
-
|
184 |
-
return [
|
185 |
-
datasets.SplitGenerator(name=datasets.Split.TRAIN, gen_kwargs={"filepath": downloaded_files["train"]}),
|
186 |
-
datasets.SplitGenerator(name=datasets.Split.VALIDATION, gen_kwargs={"filepath": downloaded_files["dev"]}),
|
187 |
-
datasets.SplitGenerator(name=datasets.Split.TEST, gen_kwargs={"filepath": downloaded_files["test"]}),
|
188 |
-
]
|
189 |
-
|
190 |
-
def _generate_examples(self, filepath):
|
191 |
-
logging.info("⏳ Generating examples from = %s", filepath)
|
192 |
-
with open(filepath, encoding="utf-8") as f:
|
193 |
-
guid = 0
|
194 |
-
tokens = []
|
195 |
-
pos_tags = []
|
196 |
-
chunk_tags = []
|
197 |
-
ner_tags = []
|
198 |
-
for line in f:
|
199 |
-
if line.startswith("-DOCSTART-") or line == "" or line == "\n":
|
200 |
-
if tokens:
|
201 |
-
yield guid, {
|
202 |
-
"id": str(guid),
|
203 |
-
"tokens": tokens,
|
204 |
-
"pos_tags": pos_tags,
|
205 |
-
"chunk_tags": chunk_tags,
|
206 |
-
"ner_tags": ner_tags,
|
207 |
-
}
|
208 |
-
guid += 1
|
209 |
-
tokens = []
|
210 |
-
pos_tags = []
|
211 |
-
chunk_tags = []
|
212 |
-
ner_tags = []
|
213 |
-
else:
|
214 |
-
# conll2003 tokens are space separated
|
215 |
-
splits = line.split(" ")
|
216 |
-
tokens.append(splits[0])
|
217 |
-
pos_tags.append(splits[1])
|
218 |
-
chunk_tags.append(splits[2])
|
219 |
-
ner_tags.append(splits[3].rstrip())
|
220 |
-
# last example
|
221 |
-
if tokens:
|
222 |
-
yield guid, {
|
223 |
-
"id": str(guid),
|
224 |
-
"tokens": tokens,
|
225 |
-
"pos_tags": pos_tags,
|
226 |
-
"chunk_tags": chunk_tags,
|
227 |
-
"ner_tags": ner_tags,
|
228 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|