Albert Sawczyn commited on
Commit
e35dccf
1 Parent(s): 552bea8

update script

Browse files
Files changed (1) hide show
  1. aspectemo.py +86 -72
aspectemo.py CHANGED
@@ -1,104 +1,118 @@
1
- import csv
2
- from typing import List, Generator, Tuple, Dict
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
3
 
4
  import datasets
5
  from datasets import DownloadManager
6
  from datasets.info import SupervisedKeysData
7
 
8
- _DESCRIPTION = """AspectEmo 1.0 dataset: Multi-Domain Corpus of Consumer Reviews for Aspect-Based
 
 
 
 
 
 
 
 
 
9
  Sentiment Analysis"""
10
 
11
- _CLASSES = ['O',
12
- 'B-a_plus_m',
13
- 'B-a_minus_m',
14
- 'B-a_zero',
15
- 'B-a_minus_s',
16
- 'B-a_plus_s',
17
- 'B-a_amb',
18
- 'B-a_minus_m:B-a_minus_m',
19
- 'B-a_minus_m:B-a_minus_m:B-a_minus_m',
20
- 'B-a_plus_m:B-a_plus_m',
21
- 'B-a_plus_m:B-a_plus_m:B-a_plus_m',
22
- 'B-a_zero:B-a_zero:B-a_zero',
23
- 'B-a_zero:B-a_zero',
24
- 'I-a_plus_m',
25
- 'B-a_zero:B-a_plus_m',
26
- 'B-a_minus_m:B-a_zero',
27
- 'B-a_minus_s:B-a_minus_s:B-a_minus_s',
28
- 'B-a_amb:B-a_amb',
29
- 'I-a_minus_m',
30
- 'B-a_minus_s:B-a_minus_s',
31
- 'B-a_plus_s:B-a_plus_s:B-a_plus_s',
32
- 'B-a_plus_m:B-a_plus_m:B-a_plus_m:B-a_plus_m:B-a_plus_m:B-a_plus_m',
33
- 'B-a_plus_m:B-a_amb',
34
- 'B-a_minus_m:B-a_plus_m',
35
- 'B-a_amb:B-a_amb:B-a_amb',
36
- 'I-a_zero',
37
- 'B-a_plus_s:B-a_plus_s',
38
- 'B-a_plus_m:B-a_plus_s',
39
- 'B-a_plus_m:B-a_zero',
40
- 'B-a_zero:B-a_zero:B-a_zero:B-a_zero:B-a_zero:B-a_zero',
41
- 'B-a_zero:B-a_minus_m',
42
- 'B-a_amb:B-a_plus_s',
43
- 'B-a_zero:B-a_minus_s']
44
 
45
- _URLS = {
46
- "train": "https://huggingface.co/datasets/clarin-pl/aspectemo/resolve/main/data/train.tsv",
47
- "validation": "https://huggingface.co/datasets/clarin-pl/aspectemo/resolve/main/data/val.tsv",
48
- "test": "https://huggingface.co/datasets/clarin-pl/aspectemo/resolve/main/data/test.tsv",
49
  }
50
 
 
 
51
 
52
  class AspectEmo(datasets.GeneratorBasedBuilder):
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
53
  def _info(self) -> datasets.DatasetInfo:
54
  return datasets.DatasetInfo(
55
  description=_DESCRIPTION,
56
  features=datasets.Features(
57
  {
58
- "orth": datasets.Sequence(datasets.Value("string")),
59
- "ctag": datasets.Sequence(datasets.Value("string")),
60
- "sentiment": datasets.Sequence(datasets.features.ClassLabel(
61
- names=_CLASSES,
62
- num_classes=len(_CLASSES)
63
- )),
64
  }
65
  ),
66
- supervised_keys=SupervisedKeysData(input="orth", output="sentiment"),
67
- homepage="https://clarin-pl.eu/dspace/handle/11321/849",
 
 
68
  )
69
 
70
- def _split_generators(self, dl_manager: DownloadManager) -> List[datasets.SplitGenerator]:
71
- urls_to_download = _URLS
72
- downloaded_files = dl_manager.download_and_extract(urls_to_download)
 
 
 
73
  return [
74
  datasets.SplitGenerator(
75
  name=datasets.Split.TRAIN,
76
- gen_kwargs={"filepath": downloaded_files["train"]},
77
- ),
78
- datasets.SplitGenerator(
79
- name=datasets.Split.VALIDATION,
80
- gen_kwargs={"filepath": downloaded_files["validation"]},
81
  ),
82
  datasets.SplitGenerator(
83
  name=datasets.Split.TEST,
84
- gen_kwargs={"filepath": downloaded_files["test"]},
 
 
 
85
  ),
86
  ]
87
 
88
  def _generate_examples(
89
- self, filepath: str
 
 
90
  ) -> Generator[Tuple[int, Dict[str, str]], None, None]:
91
- with open(filepath, "r", encoding="utf-8") as f:
92
- reader = csv.reader(f, delimiter="\t", quoting=csv.QUOTE_NONE)
93
- next(reader, None) # skip header
94
- id_, orth, ctag, sentiment = set(), [], [], []
95
- for line in reader:
96
- if not line:
97
- assert len(id_) == 1
98
- yield id_.pop(), {"orth": orth, "ctag": ctag, "sentiment": sentiment, }
99
- id_, orth, ctag, sentiment = set(), [], [], []
100
- else:
101
- id_.add(line[0])
102
- orth.append(line[1])
103
- ctag.append(line[2])
104
- sentiment.append(line[3])
 
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
+ import json
16
+ import os
17
+ from typing import Generator, Tuple, Dict, List
18
 
19
  import datasets
20
  from datasets import DownloadManager
21
  from datasets.info import SupervisedKeysData
22
 
23
+ _CITATION = """@misc{11321/849,
24
+ title = {{AspectEmo} 1.0: Multi-Domain Corpus of Consumer Reviews for Aspect-Based Sentiment Analysis},
25
+ author = {Koco{\'n}, Jan and Radom, Jarema and Kaczmarz-Wawryk, Ewa and Wabnic, Kamil and Zaj{\c a}czkowska, Ada and Za{\'s}ko-Zieli{\'n}ska, Monika},
26
+ url = {http://hdl.handle.net/11321/849},
27
+ note = {{CLARIN}-{PL} digital repository},
28
+ copyright = {The {MIT} License},
29
+ year = {2021}
30
+ }"""
31
+
32
+ _DESCRIPTION = """AspectEmo dataset: Multi-Domain Corpus of Consumer Reviews for Aspect-Based
33
  Sentiment Analysis"""
34
 
35
+ _HOMEPAGE = "https://clarin-pl.eu/dspace/handle/11321/849"
36
+
37
+ _LICENSE = "The MIT License"
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
38
 
39
+ _URLs = {
40
+ "1.0": "https://huggingface.co/datasets/clarin-pl/aspectemo/resolve/main/data/aspectemo1.zip",
41
+ # '2.0': "",
 
42
  }
43
 
44
+ _CLASSES = ["a_plus_s", "a_zero", "a_amb", "a_minus_m", "O", "a_minus_s", "a_plus_m"]
45
+
46
 
47
  class AspectEmo(datasets.GeneratorBasedBuilder):
48
+ VERSION = datasets.Version("1.0.0")
49
+
50
+ BUILDER_CONFIGS = [
51
+ datasets.BuilderConfig(
52
+ name="1.0",
53
+ version=VERSION,
54
+ description="AspectEmo 1.0 Corpus, used in the original paper.",
55
+ ),
56
+ # datasets.BuilderConfig(
57
+ # name="2.0",
58
+ # version=VERSION,
59
+ # description="",
60
+ # ),
61
+ ]
62
+
63
+ DEFAULT_CONFIG_NAME = "1.0"
64
+
65
  def _info(self) -> datasets.DatasetInfo:
66
  return datasets.DatasetInfo(
67
  description=_DESCRIPTION,
68
  features=datasets.Features(
69
  {
70
+ "tokens": datasets.Sequence(datasets.Value("string")),
71
+ "labels": datasets.Sequence(
72
+ datasets.features.ClassLabel(
73
+ names=_CLASSES, num_classes=len(_CLASSES)
74
+ )
75
+ ),
76
  }
77
  ),
78
+ supervised_keys=SupervisedKeysData(input="tokens", output="labels"),
79
+ homepage=_HOMEPAGE,
80
+ license=_LICENSE,
81
+ citation=_CITATION,
82
  )
83
 
84
+ def _split_generators(
85
+ self, dl_manager: DownloadManager
86
+ ) -> List[datasets.SplitGenerator]:
87
+ my_urls = _URLs[self.config.name]
88
+ data_dir = dl_manager.download_and_extract(my_urls)
89
+
90
  return [
91
  datasets.SplitGenerator(
92
  name=datasets.Split.TRAIN,
93
+ gen_kwargs={
94
+ "filepath": os.path.join(data_dir, "data.json"),
95
+ "split": "train",
96
+ },
 
97
  ),
98
  datasets.SplitGenerator(
99
  name=datasets.Split.TEST,
100
+ gen_kwargs={
101
+ "filepath": os.path.join(data_dir, "data.json"),
102
+ "split": "test",
103
+ },
104
  ),
105
  ]
106
 
107
  def _generate_examples(
108
+ self,
109
+ filepath: str,
110
+ split: str,
111
  ) -> Generator[Tuple[int, Dict[str, str]], None, None]:
112
+ with open(filepath, encoding="utf-8") as f:
113
+ data = json.load(f)[split]
114
+ for id_, row in data.items():
115
+ yield id_, {
116
+ "tokens": row["tokens"],
117
+ "labels": row["labels"],
118
+ }