Datasets:

Tasks:
Other
Modalities:
Text
Formats:
parquet
Languages:
English
ArXiv:
Libraries:
Datasets
Dask
License:
system HF staff commited on
Commit
ea62e45
0 Parent(s):

Update files from the datasets library (from 1.2.0)

Browse files

Release notes: https://github.com/huggingface/datasets/releases/tag/1.2.0

.gitattributes ADDED
@@ -0,0 +1,27 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ *.7z filter=lfs diff=lfs merge=lfs -text
2
+ *.arrow filter=lfs diff=lfs merge=lfs -text
3
+ *.bin filter=lfs diff=lfs merge=lfs -text
4
+ *.bin.* filter=lfs diff=lfs merge=lfs -text
5
+ *.bz2 filter=lfs diff=lfs merge=lfs -text
6
+ *.ftz filter=lfs diff=lfs merge=lfs -text
7
+ *.gz filter=lfs diff=lfs merge=lfs -text
8
+ *.h5 filter=lfs diff=lfs merge=lfs -text
9
+ *.joblib filter=lfs diff=lfs merge=lfs -text
10
+ *.lfs.* filter=lfs diff=lfs merge=lfs -text
11
+ *.model filter=lfs diff=lfs merge=lfs -text
12
+ *.msgpack filter=lfs diff=lfs merge=lfs -text
13
+ *.onnx filter=lfs diff=lfs merge=lfs -text
14
+ *.ot filter=lfs diff=lfs merge=lfs -text
15
+ *.parquet filter=lfs diff=lfs merge=lfs -text
16
+ *.pb filter=lfs diff=lfs merge=lfs -text
17
+ *.pt filter=lfs diff=lfs merge=lfs -text
18
+ *.pth filter=lfs diff=lfs merge=lfs -text
19
+ *.rar filter=lfs diff=lfs merge=lfs -text
20
+ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
21
+ *.tar.* filter=lfs diff=lfs merge=lfs -text
22
+ *.tflite filter=lfs diff=lfs merge=lfs -text
23
+ *.tgz filter=lfs diff=lfs merge=lfs -text
24
+ *.xz filter=lfs diff=lfs merge=lfs -text
25
+ *.zip filter=lfs diff=lfs merge=lfs -text
26
+ *.zstandard filter=lfs diff=lfs merge=lfs -text
27
+ *tfevents* filter=lfs diff=lfs merge=lfs -text
README.md ADDED
@@ -0,0 +1,195 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ annotations_creators:
3
+ - machine-generated
4
+ language_creators:
5
+ - found
6
+ languages:
7
+ - en
8
+ licenses:
9
+ - cc-by-4-0
10
+ multilinguality:
11
+ - monolingual
12
+ size_categories:
13
+ - n>1M
14
+ source_datasets:
15
+ - original
16
+ task_categories:
17
+ - other
18
+ task_ids:
19
+ - other-other-knowledge-base
20
+ ---
21
+
22
+ # Dataset Card for Generics KB
23
+
24
+ ## Table of Contents
25
+ - [Dataset Description](#dataset-description)
26
+ - [Dataset Summary](#dataset-summary)
27
+ - [Supported Tasks](#supported-tasks-and-leaderboards)
28
+ - [Languages](#languages)
29
+ - [Dataset Structure](#dataset-structure)
30
+ - [Data Instances](#data-instances)
31
+ - [Data Fields](#data-fields)
32
+ - [Data Splits](#data-splits)
33
+ - [Dataset Creation](#dataset-creation)
34
+ - [Curation Rationale](#curation-rationale)
35
+ - [Source Data](#source-data)
36
+ - [Annotations](#annotations)
37
+ - [Personal and Sensitive Information](#personal-and-sensitive-information)
38
+ - [Considerations for Using the Data](#considerations-for-using-the-data)
39
+ - [Social Impact of Dataset](#social-impact-of-dataset)
40
+ - [Discussion of Biases](#discussion-of-biases)
41
+ - [Other Known Limitations](#other-known-limitations)
42
+ - [Additional Information](#additional-information)
43
+ - [Dataset Curators](#dataset-curators)
44
+ - [Licensing Information](#licensing-information)
45
+ - [Citation Information](#citation-information)
46
+
47
+ ## Dataset Description
48
+
49
+ - **Homepage:** [Homepage](https://allenai.org/data/genericskb)
50
+ - **Repository:** [Repository](https://drive.google.com/drive/folders/1vqfVXhJXJWuiiXbUa4rZjOgQoJvwZUoT)
51
+ - **Paper:** [Paper](https://arxiv.org/pdf/2005.00660.pdf)
52
+ - **Point of Contact:**[Sumithra Bhakthavatsalam](sumithrab@allenai.org)
53
+ [Chloe Anastasiades](chloea@allenai.org)
54
+ [Peter Clark](peterc@allenai.org)
55
+ Alternatively email_at info@allenai.org
56
+
57
+
58
+ ### Dataset Summary
59
+
60
+ Dataset contains a large (3.5M+ sentence) knowledge base of *generic sentences*. This is the first large resource to contain *naturally occurring* generic sentences, rich in high-quality, general, semantically complete statements. All GenericsKB sentences are annotated with their topical term, surrounding context (sentences), and a (learned) confidence. We also release GenericsKB-Best (1M+ sentences), containing the best-quality generics in GenericsKB augmented with selected, synthesized generics from WordNet and ConceptNet. This demonstrates that GenericsKB can be a useful resource for NLP applications, as well as providing data for linguistic studies of generics and their semantics.
61
+
62
+ ### Supported Tasks and Leaderboards
63
+
64
+ [More Information Needed]
65
+
66
+ ### Languages
67
+
68
+ The dataset is in English.
69
+
70
+ ## Dataset Structure
71
+
72
+ ### Data Instances
73
+
74
+ The GENERICSKB contains 3,433,000 sentences. GENERICS-KB-BEST comprises of GENERICSKB generics with a score > 0.234, augmented with short generics synthesized from three other resources for all the terms (generic categories) in GENERICSKB- BEST. GENERICSKB-BEST contains 1,020,868 generics (774,621 from GENERICSKB plus 246,247 synthesized).
75
+ SimpleWikipedia is a filtered scrape of SimpleWikipedia pages (simple.wikipedia.org). The Waterloo corpus is 280GB of English plain text, gathered by Charles Clarke (Univ. Waterloo) using a webcrawler in 2001 from .edu domains.
76
+
77
+ ###### Sample SimpleWikipedia/ Waterloo config look like this
78
+ ```
79
+ {'source_name': 'SimpleWikipedia', 'sentence': 'Sepsis happens when the bacterium enters the blood and make it form tiny clots.', 'sentences_before': [], 'sentences_after': [], 'concept_name': 'sepsis', 'quantifiers': {}, 'id': 'SimpleWikipedia--tmp-sw-rs1-with-bug-fixes-initialprocessing-inputs-articles-with-clean-sentences-jsonl-c27816b298e1e0b5326916ee4e2fd0f1603caa77-100-Bubonic-plague--Different-kinds-of-the-same-disease--Septicemic-plague-0-0-039fbe9c11adde4ff9a829376ca7e0ed-1560874903-47882-/Users/chloea/Documents/aristo/commonsense/kbs/simplewikipedia/commonsense-filtered-good-rs1.jsonl-1f33b1e84018a2b1bfdf446f9a6491568b5585da-1561086091.8220549', 'bert_score': 0.8396177887916565}
80
+ ```
81
+ ###### Sample instance for Generics KB datasets look like this:
82
+ ```
83
+ {'source': 'Waterloo', 'term': 'aardvark', 'quantifier_frequency': '', 'quantifier_number': '', 'generic_sentence': 'Aardvarks are very gentle animals.', 'score': '0.36080607771873474'}
84
+ {'source': 'TupleKB', 'term': 'aardvark', 'quantifier_frequency': '', 'quantifier_number': '', 'generic_sentence': 'Aardvarks dig burrows.', 'score': '1.0'}
85
+ ```
86
+ ### Data Fields
87
+
88
+ The fields in GenericsKB-Best.tsv and GenericsKB.tsv are as follows:
89
+ - `SOURCE`: denotes the source of the generic
90
+ - `TERM`: denotes the category that is the topic of the generic.
91
+ - `GENERIC SENTENCE`: is the sentence itself.
92
+ - `SCORE`: Is the BERT-trained score, measuring the degree to which the generic represents a "useful, general truth" about the world (as judged by crowdworkers). Score ranges from 0 (worst) to 1 (best). Sentences with scores below 0.23 (corresponding to an "unsure" vote by crowdworkers) are in GenericsKB, but are not part of GenericsKB-Best due to their unreliability.
93
+ - `QUANTIFIER_FREQUENCY`:For generics with explicit quantifiers (all, most, etc.) the quantifier is listed - Frequency contains values such as 'usually', 'often', 'frequently'
94
+ - `QUANTIFIER_NUMBER`: For generics with explicit quantifiers (all, most, etc.) with values such as 'all'|'any'|'most'|'much'|'some' etc...
95
+
96
+ The SimpleWiki/Waterloo generics from GenericsKB.tsv, but expanded to also include their surrounding context (before/after sentences). The Waterloo generics are the majority of GenericsKB. This zip file is 1.4GB expanding to 5.5GB.
97
+ There is a json representation for every generic statement in the Generics KB. The generic statement is stored under the `sentence` field within the `knowledge` object. There is also a `bert_score` associated with each sentence which is the BERT-based classifier's score for the 'genericness' of the statement. This score is meant to reflect how much generalized world knowledge/commonsense the statement captures vs only being contextually meaningful.
98
+ Detailed description of each of the fields:
99
+
100
+ - `source_name`: The name of the corpus the generic statement was picked from.
101
+ - `sentence`: The generic sentence.
102
+ - `sentences_before`: Provides context information surrounding the generic statement from the original corpus.Up to five sentences preceding the generic sentence in the original corpus.
103
+ - `sentences_after`: Up to five sentences following the generic sentence in the original corpus.
104
+ - `concept_name`: A concept that is the subject of the generic statement.
105
+ - `quantifiers`: The quantifiers for the key concept of the generic statement. There can be multiple quantifiers to allow for statements such as "All bats sometimes fly", where 'all' and 'sometimes' are both quantifiers reflecting number and frequency respectively.
106
+ - `id`: Unique identifier for a generic statement in the kb.
107
+ - `bert_score`: Score for the generic statement from the BERT-based generics classifier.
108
+ <br>**Additional fields that apply only to SimpleWiki dataset**
109
+ - `headings`: A breadcrumb of section/subsection headings from the top down to the location of the generic statement in the corpus. It applies to SimpleWikipedia which has a hierarchical structure.
110
+ - `categories`:The listed categories under which the source article falls. Applies to SimpleWikipedia.
111
+
112
+
113
+ ### Data Splits
114
+
115
+ There are no splits.
116
+
117
+ ## Dataset Creation
118
+
119
+ ### Curation Rationale
120
+
121
+ [More Information Needed]
122
+
123
+ ### Source Data
124
+
125
+ #### Initial Data Collection and Normalization
126
+
127
+ Data was crawled. SimpleWikipedia is a filtered scrape of SimpleWikipedia pages (simple.wikipedia.org). The Waterloo corpus is 280GB of English plain text, gathered by Charles Clarke (Univ. Waterloo) using a webcrawler in 2001 from .edu domains.
128
+
129
+ #### Who are the source language producers?
130
+
131
+ [More Information Needed]
132
+
133
+ ### Annotations
134
+
135
+ #### Annotation process
136
+
137
+ Bert was used to decide whether the sentence is useful or not. Every sentence has a bert score.
138
+
139
+ #### Who are the annotators?
140
+
141
+ No annotations were made.
142
+
143
+ ### Personal and Sensitive Information
144
+
145
+ [More Information Needed]
146
+
147
+ ## Considerations for Using the Data
148
+
149
+ ### Social Impact of Dataset
150
+
151
+ [More Information Needed]
152
+
153
+ ### Discussion of Biases
154
+
155
+ [More Information Needed]
156
+
157
+ ### Other Known Limitations
158
+
159
+ [More Information Needed]
160
+
161
+ ## Additional Information
162
+
163
+ ### Dataset Curators
164
+
165
+ [More Information Needed]
166
+
167
+ ### Licensing Information
168
+
169
+ The GenericsKB is available under the Creative Commons - Attribution 4.0 International - licence.
170
+
171
+ As an informal summary, from https://creativecommons.org/licenses/by/4.0/, you are free to:
172
+
173
+ Share ― copy and redistribute the material in any medium or format
174
+ Adapt ― remix, transform, and build upon the material for any purpose, even commercially.
175
+
176
+ under the following terms:
177
+
178
+ Attribution ― You must give appropriate credit, provide a link to the license, and
179
+ indicate if changes were made. You may do so in any reasonable manner,
180
+ but not in any way that suggests the licensor endorses you or your use.
181
+ No additional restrictions ― You may not apply legal terms or technological measures
182
+ that legally restrict others from doing anything the license permits.
183
+
184
+ For details, see https://creativecommons.org/licenses/by/4.0/ or the or the included
185
+ file "Creative Commons ― Attribution 4.0 International ― CC BY 4.0.pdf" in this folder.
186
+
187
+ ### Citation Information
188
+ ```
189
+ @InProceedings{huggingface:dataset,
190
+ title = {GenericsKB: A Knowledge Base of Generic Statements},
191
+ authors={Sumithra Bhakthavatsalam, Chloe Anastasiades, Peter Clark},
192
+ year={2020},
193
+ publisher = {Allen Institute for AI},
194
+ }
195
+ ```
dataset_infos.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"generics_kb_best": {"description": "The GenericsKB contains 3.4M+ generic sentences about the world, i.e., sentences expressing general truths such as \"Dogs bark,\" and \"Trees remove carbon dioxide from the atmosphere.\" Generics are potentially useful as a knowledge source for AI systems requiring general world knowledge. The GenericsKB is the first large-scale resource containing naturally occurring generic sentences (as opposed to extracted or crowdsourced triples), and is rich in high-quality, general, semantically complete statements. Generics were primarily extracted from three large text sources, namely the Waterloo Corpus, selected parts of Simple Wikipedia, and the ARC Corpus. A filtered, high-quality subset is also available in GenericsKB-Best, containing 1,020,868 sentences. We recommend you start with GenericsKB-Best.\n", "citation": "@InProceedings{huggingface:dataset,\ntitle = {GenericsKB: A Knowledge Base of Generic Statements},\nauthors={Sumithra Bhakthavatsalam, Chloe Anastasiades, Peter Clark},\nyear={2020},\npublisher = {Allen Institute for AI},\n}\n", "homepage": "https://allenai.org/data/genericskb", "license": "cc-by-4.0", "features": {"source": {"dtype": "string", "id": null, "_type": "Value"}, "term": {"dtype": "string", "id": null, "_type": "Value"}, "quantifier_frequency": {"dtype": "string", "id": null, "_type": "Value"}, "quantifier_number": {"dtype": "string", "id": null, "_type": "Value"}, "generic_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "score": {"dtype": "float64", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "builder_name": "generics_kb", "config_name": "generics_kb_best", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 99897719, "num_examples": 1020868, "dataset_name": "generics_kb"}}, "download_checksums": {"https://drive.google.com/u/0/uc?id=12DfIzoWyHIQqssgUgDvz3VG8_ScSh6ng&export=download": {"num_bytes": 94850525, "checksum": "0668b23c8b1579b6a76fcf48e04f3c9ea039ca9048a26848151d689deabb75e2"}}, "download_size": 94850525, "post_processing_size": null, "dataset_size": 99897719, "size_in_bytes": 194748244}, "generics_kb": {"description": "The GenericsKB contains 3.4M+ generic sentences about the world, i.e., sentences expressing general truths such as \"Dogs bark,\" and \"Trees remove carbon dioxide from the atmosphere.\" Generics are potentially useful as a knowledge source for AI systems requiring general world knowledge. The GenericsKB is the first large-scale resource containing naturally occurring generic sentences (as opposed to extracted or crowdsourced triples), and is rich in high-quality, general, semantically complete statements. Generics were primarily extracted from three large text sources, namely the Waterloo Corpus, selected parts of Simple Wikipedia, and the ARC Corpus. A filtered, high-quality subset is also available in GenericsKB-Best, containing 1,020,868 sentences. We recommend you start with GenericsKB-Best.\n", "citation": "@InProceedings{huggingface:dataset,\ntitle = {GenericsKB: A Knowledge Base of Generic Statements},\nauthors={Sumithra Bhakthavatsalam, Chloe Anastasiades, Peter Clark},\nyear={2020},\npublisher = {Allen Institute for AI},\n}\n", "homepage": "https://allenai.org/data/genericskb", "license": "cc-by-4.0", "features": {"source": {"dtype": "string", "id": null, "_type": "Value"}, "term": {"dtype": "string", "id": null, "_type": "Value"}, "quantifier_frequency": {"dtype": "string", "id": null, "_type": "Value"}, "quantifier_number": {"dtype": "string", "id": null, "_type": "Value"}, "generic_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "score": {"dtype": "float64", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "builder_name": "generics_kb", "config_name": "generics_kb", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 348158966, "num_examples": 3433000, "dataset_name": "generics_kb"}}, "download_checksums": {"https://drive.google.com/u/0/uc?id=1UOIEzQTid7SzKx2tbwSSPxl7g-CjpoZa&export=download": {"num_bytes": 343284785, "checksum": "7ec2419e700b3425129032f75f0bb01887bdb84231526468751d6cc2a9b9e61e"}}, "download_size": 343284785, "post_processing_size": null, "dataset_size": 348158966, "size_in_bytes": 691443751}, "generics_kb_simplewiki": {"description": "The GenericsKB contains 3.4M+ generic sentences about the world, i.e., sentences expressing general truths such as \"Dogs bark,\" and \"Trees remove carbon dioxide from the atmosphere.\" Generics are potentially useful as a knowledge source for AI systems requiring general world knowledge. The GenericsKB is the first large-scale resource containing naturally occurring generic sentences (as opposed to extracted or crowdsourced triples), and is rich in high-quality, general, semantically complete statements. Generics were primarily extracted from three large text sources, namely the Waterloo Corpus, selected parts of Simple Wikipedia, and the ARC Corpus. A filtered, high-quality subset is also available in GenericsKB-Best, containing 1,020,868 sentences. We recommend you start with GenericsKB-Best.\n", "citation": "@InProceedings{huggingface:dataset,\ntitle = {GenericsKB: A Knowledge Base of Generic Statements},\nauthors={Sumithra Bhakthavatsalam, Chloe Anastasiades, Peter Clark},\nyear={2020},\npublisher = {Allen Institute for AI},\n}\n", "homepage": "https://allenai.org/data/genericskb", "license": "cc-by-4.0", "features": {"source_name": {"dtype": "string", "id": null, "_type": "Value"}, "sentence": {"dtype": "string", "id": null, "_type": "Value"}, "sentences_before": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "sentences_after": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "concept_name": {"dtype": "string", "id": null, "_type": "Value"}, "quantifiers": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "id": {"dtype": "string", "id": null, "_type": "Value"}, "bert_score": {"dtype": "float64", "id": null, "_type": "Value"}, "headings": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "categories": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "builder_name": "generics_kb", "config_name": "generics_kb_simplewiki", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 10039355, "num_examples": 12765, "dataset_name": "generics_kb"}}, "download_checksums": {"https://drive.google.com/u/0/uc?id=1SpN9Qc7XRy5xs4tIfXkcLOEAP2IVaK15&export=download": {"num_bytes": 16437369, "checksum": "f6c0da9c9100172e8979907448497717d8ea1a50ee96aa2b81e447423c6cd0bb"}}, "download_size": 16437369, "post_processing_size": null, "dataset_size": 10039355, "size_in_bytes": 26476724}, "generics_kb_waterloo": {"description": "The GenericsKB contains 3.4M+ generic sentences about the world, i.e., sentences expressing general truths such as \"Dogs bark,\" and \"Trees remove carbon dioxide from the atmosphere.\" Generics are potentially useful as a knowledge source for AI systems requiring general world knowledge. The GenericsKB is the first large-scale resource containing naturally occurring generic sentences (as opposed to extracted or crowdsourced triples), and is rich in high-quality, general, semantically complete statements. Generics were primarily extracted from three large text sources, namely the Waterloo Corpus, selected parts of Simple Wikipedia, and the ARC Corpus. A filtered, high-quality subset is also available in GenericsKB-Best, containing 1,020,868 sentences. We recommend you start with GenericsKB-Best.\n", "citation": "@InProceedings{huggingface:dataset,\ntitle = {GenericsKB: A Knowledge Base of Generic Statements},\nauthors={Sumithra Bhakthavatsalam, Chloe Anastasiades, Peter Clark},\nyear={2020},\npublisher = {Allen Institute for AI},\n}\n", "homepage": "https://allenai.org/data/genericskb", "license": "cc-by-4.0", "features": {"source_name": {"dtype": "string", "id": null, "_type": "Value"}, "sentence": {"dtype": "string", "id": null, "_type": "Value"}, "sentences_before": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "sentences_after": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "concept_name": {"dtype": "string", "id": null, "_type": "Value"}, "quantifiers": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "id": {"dtype": "string", "id": null, "_type": "Value"}, "bert_score": {"dtype": "float64", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "builder_name": "generics_kb", "config_name": "generics_kb_waterloo", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 4277214701, "num_examples": 3666725, "dataset_name": "generics_kb"}}, "download_checksums": {}, "download_size": 0, "post_processing_size": null, "dataset_size": 4277214701, "size_in_bytes": 4277214701}}
dummy/generics_kb/1.0.0/dummy_data.zip ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7b32717594e3df07c6365c073c3cbce34da9af451e3ac9e0ece22ea171d94561
3
+ size 624
dummy/generics_kb_best/1.0.0/dummy_data.zip ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:155593c88d4746c25fb1811988749c236a4a1924a59d90254130bda6b21b1d06
3
+ size 638
dummy/generics_kb_simplewiki/1.0.0/dummy_data.zip ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b1ea088a2c4e3b97d30db11cb0a48efaf6d0ad8ce48ef1b96d78d71c3148d97c
3
+ size 1828
dummy/generics_kb_waterloo/1.0.0/dummy_data.zip ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:665e5b672592d4dfa1185ebeb727d80b94707a228ffaa7e1d88a9cc34b3956c7
3
+ size 1320
generics_kb.py ADDED
@@ -0,0 +1,232 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
+ """Generics KB: A Knowledge Base of Generic Statements"""
16
+
17
+ from __future__ import absolute_import, division, print_function
18
+
19
+ import ast
20
+ import csv
21
+ import os
22
+
23
+ import datasets
24
+
25
+
26
+ # TODO: Add BibTeX citation
27
+ # Find for instance the citation on arxiv or on the dataset repo/website
28
+ _CITATION = """\
29
+ @InProceedings{huggingface:dataset,
30
+ title = {GenericsKB: A Knowledge Base of Generic Statements},
31
+ authors={Sumithra Bhakthavatsalam, Chloe Anastasiades, Peter Clark},
32
+ year={2020},
33
+ publisher = {Allen Institute for AI},
34
+ }
35
+ """
36
+
37
+ _DESCRIPTION = """\
38
+ The GenericsKB contains 3.4M+ generic sentences about the world, i.e., sentences expressing general truths such as "Dogs bark," and "Trees remove carbon dioxide from the atmosphere." Generics are potentially useful as a knowledge source for AI systems requiring general world knowledge. The GenericsKB is the first large-scale resource containing naturally occurring generic sentences (as opposed to extracted or crowdsourced triples), and is rich in high-quality, general, semantically complete statements. Generics were primarily extracted from three large text sources, namely the Waterloo Corpus, selected parts of Simple Wikipedia, and the ARC Corpus. A filtered, high-quality subset is also available in GenericsKB-Best, containing 1,020,868 sentences. We recommend you start with GenericsKB-Best.
39
+ """
40
+
41
+ _HOMEPAGE = "https://allenai.org/data/genericskb"
42
+
43
+ _LICENSE = "cc-by-4.0"
44
+
45
+ _URL = "https://drive.google.com/u/0/uc?id={0}&export=download"
46
+
47
+ _FILEPATHS = {
48
+ "generics_kb_best": _URL.format("12DfIzoWyHIQqssgUgDvz3VG8_ScSh6ng"),
49
+ "generics_kb": _URL.format("1UOIEzQTid7SzKx2tbwSSPxl7g-CjpoZa"),
50
+ "generics_kb_simplewiki": _URL.format("1SpN9Qc7XRy5xs4tIfXkcLOEAP2IVaK15"),
51
+ "generics_kb_waterloo": "cskb-waterloo-06-21-with-bert-scores.jsonl",
52
+ }
53
+
54
+
55
+ class GenericsKb(datasets.GeneratorBasedBuilder):
56
+ """ The GenericsKB is the first large-scale resource containing naturally occurring generic sentences, and is rich in high-quality, general, semantically complete statements."""
57
+
58
+ VERSION = datasets.Version("1.0.0")
59
+
60
+ BUILDER_CONFIGS = [
61
+ datasets.BuilderConfig(
62
+ name="generics_kb_best",
63
+ version=VERSION,
64
+ description="This is the default and recommended config.Comprises of GENERICSKB generics with a score > 0.234 ",
65
+ ),
66
+ datasets.BuilderConfig(
67
+ name="generics_kb", version=VERSION, description="This GENERICSKB that contains 3,433,000 sentences."
68
+ ),
69
+ datasets.BuilderConfig(
70
+ name="generics_kb_simplewiki",
71
+ version=VERSION,
72
+ description="SimpleWikipedia is a filtered scrape of SimpleWikipedia pages (simple.wikipedia.org)",
73
+ ),
74
+ datasets.BuilderConfig(
75
+ name="generics_kb_waterloo",
76
+ version=VERSION,
77
+ description="The Waterloo corpus is 280GB of English plain text, gathered by Charles Clarke (Univ. Waterloo) using a webcrawler in 2001 from .edu domains.",
78
+ ),
79
+ ]
80
+
81
+ @property
82
+ def manual_download_instructions(self):
83
+ return """\
84
+ You need to manually download the files needed for the dataset config generics_kb_waterloo. The other configs like generics_kb_best don't need manual downloads.
85
+ The <path/to/folder> can e.g. be `~/Downloads/GenericsKB`. Download the following required files from https://drive.google.com/drive/folders/1vqfVXhJXJWuiiXbUa4rZjOgQoJvwZUoT
86
+ For working on "generics_kb_waterloo" data,
87
+ 1. Manually download 'GenericsKB-Waterloo-WithContext.jsonl.zip' into your <path/to/folder>.Please ensure the filename is as is.
88
+ The Waterloo is also generics from GenericsKB.tsv, but expanded to also include their surrounding context (before/after sentences). The Waterloo generics are the majority of GenericsKB. This zip file is 1.4GB expanding to 5.5GB.
89
+ 2. Extract the GenericsKB-Waterloo-WithContext.jsonl.zip; It will create a file of 5.5 GB called cskb-waterloo-06-21-with-bert-scores.jsonl.
90
+ Ensure you move this file into your <path/to/folder>.
91
+
92
+ generics_kb can then be loaded using the following commands based on which data you want to work on. Data files must be present in the <path/to/folder> if using "generics_kb_waterloo" config.
93
+ 1. `datasets.load_dataset("generics_kb","generics_kb_best")`.
94
+ 2. `datasets.load_dataset("generics_kb","generics_kb")`
95
+ 3. `datasets.load_dataset("generics_kb","generics_kb_simplewiki")`
96
+ 4. `datasets.load_dataset("generics_kb","generics_kb_waterloo", data_dir="<path/to/folder>")`
97
+
98
+ """
99
+
100
+ DEFAULT_CONFIG_NAME = "generics_kb_best"
101
+
102
+ def _info(self):
103
+ if self.config.name == "generics_kb_waterloo" or self.config.name == "generics_kb_simplewiki":
104
+
105
+ featuredict = {
106
+ "source_name": datasets.Value("string"),
107
+ "sentence": datasets.Value("string"),
108
+ "sentences_before": datasets.Sequence(datasets.Value("string")),
109
+ "sentences_after": datasets.Sequence(datasets.Value("string")),
110
+ "concept_name": datasets.Value("string"),
111
+ "quantifiers": datasets.Sequence(datasets.Value("string")),
112
+ "id": datasets.Value("string"),
113
+ "bert_score": datasets.Value("float64"),
114
+ }
115
+ if self.config.name == "generics_kb_simplewiki":
116
+ featuredict["headings"] = datasets.Sequence(datasets.Value("string"))
117
+ featuredict["categories"] = datasets.Sequence(datasets.Value("string"))
118
+
119
+ features = datasets.Features(featuredict)
120
+
121
+ else:
122
+
123
+ features = datasets.Features(
124
+ {
125
+ "source": datasets.Value("string"),
126
+ "term": datasets.Value("string"),
127
+ "quantifier_frequency": datasets.Value("string"),
128
+ "quantifier_number": datasets.Value("string"),
129
+ "generic_sentence": datasets.Value("string"),
130
+ "score": datasets.Value("float64"),
131
+ }
132
+ )
133
+
134
+ return datasets.DatasetInfo(
135
+ # This is the description that will appear on the datasets page.
136
+ description=_DESCRIPTION,
137
+ # This defines the different columns of the dataset and their types
138
+ features=features, # Here we define them above because they are different between the two configurations
139
+ # If there's a common (input, target) tuple from the features,
140
+ # specify them here. They'll be used if as_supervised=True in
141
+ # builder.as_dataset.
142
+ supervised_keys=None,
143
+ # Homepage of the dataset for documentation
144
+ homepage=_HOMEPAGE,
145
+ # License for the dataset if available
146
+ license=_LICENSE,
147
+ # Citation for the dataset
148
+ citation=_CITATION,
149
+ )
150
+
151
+ def _split_generators(self, dl_manager):
152
+
153
+ if self.config.name == "generics_kb_waterloo":
154
+ data_dir = os.path.abspath(os.path.expanduser(dl_manager.manual_dir))
155
+ # check if manual folder exists
156
+ if not os.path.exists(data_dir):
157
+ raise FileNotFoundError(
158
+ f"{data_dir} does not exist. Make sure you insert a manual dir via `datasets.load_dataset('generics_kb', data_dir=...)`. Manual download instructions: {self.manual_download_instructions})"
159
+ )
160
+
161
+ # Check if required files exist in the folder
162
+ filepath = os.path.join(data_dir, _FILEPATHS[self.config.name])
163
+
164
+ if not os.path.exists(filepath):
165
+ raise FileNotFoundError(
166
+ f"{filepath} does not exist. Make sure you required files are present in {data_dir} `. Manual download instructions: {self.manual_download_instructions})"
167
+ )
168
+ else:
169
+ filepath = dl_manager.download(_FILEPATHS[self.config.name])
170
+
171
+ return [
172
+ datasets.SplitGenerator(
173
+ name=datasets.Split.TRAIN,
174
+ # These kwargs will be passed to _generate_examples
175
+ gen_kwargs={
176
+ "filepath": filepath,
177
+ },
178
+ ),
179
+ ]
180
+
181
+ def _generate_examples(self, filepath):
182
+ """ Yields examples. """
183
+
184
+ if self.config.name == "generics_kb_waterloo" or self.config.name == "generics_kb_simplewiki":
185
+
186
+ with open(filepath, encoding="utf-8") as f:
187
+ for id_, row in enumerate(f):
188
+ data = ast.literal_eval(row)
189
+
190
+ result = {
191
+ "source_name": data["source"]["name"],
192
+ "sentence": data["knowledge"]["sentence"],
193
+ "sentences_before": data["knowledge"]["context"]["sentences_before"],
194
+ "sentences_after": data["knowledge"]["context"]["sentences_after"],
195
+ "concept_name": data["knowledge"]["key_concepts"][0]["concept_name"],
196
+ "quantifiers": data["knowledge"]["key_concepts"][0]["quantifiers"],
197
+ "id": data["id"],
198
+ "bert_score": data["bert_score"],
199
+ }
200
+ if self.config.name == "generics_kb_simplewiki":
201
+ result["headings"] = data["knowledge"]["context"]["headings"]
202
+ result["categories"] = data["knowledge"]["context"]["categories"]
203
+
204
+ yield id_, result
205
+ else:
206
+
207
+ with open(filepath, encoding="utf-8") as f:
208
+ # Skip the header
209
+ next(f)
210
+
211
+ read_tsv = csv.reader(f, delimiter="\t")
212
+
213
+ for id_, row in enumerate(read_tsv):
214
+
215
+ quantifier = row[2]
216
+ quantifier_frequency = ""
217
+ quantifier_number = ""
218
+
219
+ if quantifier != "":
220
+ quantifier = ast.literal_eval(quantifier)
221
+ if "frequency" in quantifier.keys():
222
+ quantifier_frequency = quantifier["frequency"]
223
+ if "number" in quantifier.keys():
224
+ quantifier_number = quantifier["number"]
225
+ yield id_, {
226
+ "source": row[0],
227
+ "term": row[1],
228
+ "quantifier_frequency": quantifier_frequency,
229
+ "quantifier_number": quantifier_number,
230
+ "generic_sentence": row[3],
231
+ "score": row[4],
232
+ }