wardenga commited on
Commit
763edf7
1 Parent(s): 96270b0

initial commit

Browse files
Files changed (2) hide show
  1. dataset_infos.json +1 -0
  2. lsoie.py +151 -0
dataset_infos.json ADDED
@@ -0,0 +1 @@
 
1
+ {"wiki": {"description": "\nThe Large Scale Open Information Extraction Dataset (LSOIE), is a dataset 20 \ntimes larger than the next largest human-annotated Open Information Extraction\n(OIE) dataset. LSOIE is a built upon the QA-SRL 2.0 dataset.\n", "citation": "@article{lsoie-2021,\n title={{LSOIE}: A Large-Scale Dataset for Supervised Open Information Extraction},\n author={{Solawetz}, Jacob and {Larson}, Stefan},\n journal={arXiv preprint arXiv:2101.11177},\n year={2019},\n url=\"https://arxiv.org/pdf/2101.11177.pdf\"\n}\n", "homepage": "https://github.com/Jacobsolawetz/large-scale-oie/", "license": "", "features": {"word_ids": {"feature": {"dtype": "int16", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "words": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "pred": {"dtype": "string", "id": null, "_type": "Value"}, "pred_ids": {"feature": {"dtype": "int16", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "head_pred_id": {"dtype": "int16", "id": null, "_type": "Value"}, "sent_id": {"dtype": "int16", "id": null, "_type": "Value"}, "run_id": {"dtype": "int16", "id": null, "_type": "Value"}, "label": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": {"input": "word_ids", "output": "label"}, "task_templates": null, "builder_name": "lsoie", "config_name": "wiki", "version": "0.0.0", "splits": {"train": {"name": "train", "num_bytes": 24938522, "num_examples": 46016, "dataset_name": "lsoie"}, "validation": {"name": "validation", "num_bytes": 2880854, "num_examples": 5269, "dataset_name": "lsoie"}, "test": {"name": "test", "num_bytes": 2840517, "num_examples": 5374, "dataset_name": "lsoie"}}, "download_checksums": {"https://github.com/Jacobsolawetz/large-scale-oie/raw/master/dataset_creation/lsoie_data/lsoie_data.zip": {"num_bytes": 19799926, "checksum": "0d189a3a8fef4b9f9efdad8faf0f53fc53805f9b2ad5354926e09c1449a00330"}}, "download_size": 19799926, "post_processing_size": null, "dataset_size": 30659893, "size_in_bytes": 50459819}}
lsoie.py ADDED
@@ -0,0 +1,151 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # -*- coding: utf-8 -*-
2
+ """LSOIE: A Large-Scale Dataset for Supervised Open Information Extraction."""
3
+ import os
4
+ import datasets
5
+ from datasets.info import SupervisedKeysData
6
+ from zipfile import ZipFile
7
+ logger = datasets.logging.get_logger(__name__)
8
+
9
+
10
+ _CITATION = """\
11
+ @article{lsoie-2021,
12
+ title={{LSOIE}: A Large-Scale Dataset for Supervised Open Information Extraction},
13
+ author={{Solawetz}, Jacob and {Larson}, Stefan},
14
+ journal={arXiv preprint arXiv:2101.11177},
15
+ year={2019},
16
+ url="https://arxiv.org/pdf/2101.11177.pdf"
17
+ }
18
+ """
19
+
20
+ _DESCRIPTION = """
21
+ The Large Scale Open Information Extraction Dataset (LSOIE), is a dataset 20
22
+ times larger than the next largest human-annotated Open Information Extraction
23
+ (OIE) dataset. LSOIE is a built upon the QA-SRL 2.0 dataset.
24
+ """
25
+
26
+ _URL = "https://github.com/Jacobsolawetz/large-scale-oie/"
27
+ _URLS = {
28
+ "zip": _URL+"raw/master/dataset_creation/lsoie_data/lsoie_data.zip"
29
+ }
30
+ _ARCHIVE_FILES = [
31
+ "lsoie_science_train.conll",
32
+ "lsoie_science_dev.conll",
33
+ "lsoie_science_test.conll",
34
+ "lsoie_wiki_train.conll",
35
+ "lsoie_wiki_dev.conll",
36
+ "lsoie_wiki_test.conll",
37
+ ]
38
+
39
+
40
+ class LsoieConfig(datasets.BuilderConfig):
41
+ """BuilderConfig for LSOIE."""
42
+
43
+ def __init__(self,subset="wiki", **kwargs):
44
+ """BuilderConfig for LSOIE.
45
+ Args:
46
+ subset: str - either "wiki" or "science"
47
+ **kwargs: keyword arguments forwarded to super.
48
+ """
49
+ super(LsoieConfig, self).__init__(**kwargs)
50
+ self.subset=subset
51
+
52
+
53
+ class Lsoie(datasets.GeneratorBasedBuilder):
54
+ """LSOIE: A Large-Scale Dataset for Supervised Open Information Extraction"""
55
+
56
+ BUILDER_CONFIGS = [
57
+ LsoieConfig(
58
+ name="wiki",
59
+ description="LSOIE dataset from wikipedia and wikinews",
60
+ subset="wiki",
61
+ ),
62
+ LsoieConfig(
63
+ name="sci",
64
+ description="LSOIE dataset build over scientific domain",
65
+ subset="sci",
66
+ ),
67
+ ]
68
+
69
+ DEFAULT_CONFIG_NAME = "wiki"
70
+
71
+ def _info(self):
72
+ return datasets.DatasetInfo(
73
+ description=_DESCRIPTION,
74
+ features=datasets.Features(
75
+ {
76
+ "word_ids": datasets.Sequence(datasets.Value("int16")),
77
+ "words": datasets.Sequence(datasets.Value("string")),
78
+ "pred": datasets.Value("string"),
79
+ "pred_ids": datasets.Sequence(datasets.Value("int16")),
80
+ "head_pred_id": datasets.Value("int16"),
81
+ "sent_id": datasets.Value("int16"),
82
+ "run_id": datasets.Value("int16"),
83
+ "label": datasets.Sequence(datasets.Value("string")),
84
+ }
85
+ ),
86
+ supervised_keys=SupervisedKeysData(input="word_ids",output="label"),
87
+ homepage=_URL,
88
+ citation=_CITATION,
89
+ #there is no default task for open information extraction yet
90
+ #task_templates=[
91
+ # OpenInformationExtraction(
92
+ # question_column="question", context_column="context", answers_column="answers"
93
+ # )
94
+ #],
95
+ )
96
+
97
+ def _split_generators(self, dl_manager):
98
+ downloaded_archive = dl_manager.download(_URLS)['zip']
99
+ #name_pre=os.path.join("lsoie_data","lsoie_")+self.config.subset+"_"
100
+ name_pre="lsoie_"+self.config.subset+"_"
101
+ return [
102
+ datasets.SplitGenerator(name=datasets.Split.TRAIN,
103
+ gen_kwargs={
104
+ "archive_path": downloaded_archive,
105
+ "file_name": name_pre+"train.conll",
106
+ }),
107
+ datasets.SplitGenerator(name=datasets.Split.VALIDATION,
108
+ gen_kwargs={
109
+ "archive_path": downloaded_archive,
110
+ "file_name": name_pre+"dev.conll",
111
+ }),
112
+ datasets.SplitGenerator(name=datasets.Split.TEST,
113
+ gen_kwargs={
114
+ "archive_path": downloaded_archive,
115
+ "file_name": name_pre+"test.conll",
116
+ }),
117
+ ]
118
+
119
+ def _generate_examples(self,archive_path,file_name):
120
+ """This functions returns the samples in a raw format"""
121
+ logger.info("generating examples from archive:{}".format(archive_path))
122
+ columns={'word_ids':int,
123
+ 'words':str,
124
+ 'pred':str,
125
+ 'pred_ids':lambda x: [ num for num in x.strip('[]').split(',')],
126
+ 'head_pred_id': int,
127
+ 'sent_id':int,
128
+ 'run_id': int,
129
+ 'label':str}
130
+ list_columns=["word_ids","words","label"]
131
+ sep="\t"
132
+ key=0
133
+ sentence=dict()
134
+ for column in list_columns:
135
+ sentence[column]=[]
136
+ with ZipFile(archive_path) as zipfile:
137
+ with zipfile.open('lsoie_data/'+file_name,mode='r') as file:
138
+ for line in file:
139
+ line=line.decode("utf-8").strip('\n').split(sep=sep)
140
+ if line[0]=='':
141
+ yield key, sentence
142
+ key+=1
143
+ for column in list_columns:
144
+ sentence[column]=[]
145
+ continue
146
+ for column, val in zip(columns.keys(),line):
147
+ val=columns[column](val)
148
+ if column in list_columns:
149
+ sentence[column].append(val)
150
+ else:
151
+ sentence[column]=val