Datasets:
Tasks:
Text2Text Generation
Sub-tasks:
text-simplification
Languages:
English
Size:
100K<n<1M
ArXiv:
License:
Update files from the datasets library (from 1.2.1)
Browse filesRelease notes: https://github.com/huggingface/datasets/releases/tag/1.2.1
- README.md +6 -2
- dataset_infos.json +1 -1
- dummy/auto_full_no_split/1.0.0/dummy_data.zip +3 -0
- dummy/auto_full_with_split/1.0.0/dummy_data.zip +3 -0
- wiki_auto.py +32 -4
README.md
CHANGED
@@ -4,6 +4,10 @@ annotations_creators:
|
|
4 |
- machine-generated
|
5 |
auto_acl:
|
6 |
- machine-generated
|
|
|
|
|
|
|
|
|
7 |
manual:
|
8 |
- crowdsourced
|
9 |
language_creators:
|
@@ -61,7 +65,7 @@ WikiAuto provides a set of aligned sentences from English Wikipedia and Simple E
|
|
61 |
|
62 |
The authors first crowd-sourced a set of manual alignments between sentences in a subset of the Simple English Wikipedia and their corresponding versions in English Wikipedia (this corresponds to the `manual` config in this version of dataset), then trained a neural CRF system to predict these alignments.
|
63 |
|
64 |
-
The trained alignment prediction model was then applied to the other articles in Simple English Wikipedia with an English counterpart to create a larger corpus of aligned sentences (corresponding to the `auto` and `
|
65 |
|
66 |
### Supported Tasks and Leaderboards
|
67 |
|
@@ -127,7 +131,7 @@ The `auto` config shows a pair of an English and corresponding Simple English Wi
|
|
127 |
'simple_article_url': 'https://simple.wikipedia.org/wiki?curid=702227'}}
|
128 |
```
|
129 |
|
130 |
-
Finally, the `auto_acl`
|
131 |
```
|
132 |
{'normal_sentence': 'In early work , Rutherford discovered the concept of radioactive half-life , the radioactive element radon , and differentiated and named alpha and beta radiation .\n',
|
133 |
'simple_sentence': 'Rutherford discovered the radioactive half-life , and the three parts of radiation which he named Alpha , Beta , and Gamma .\n'}
|
|
|
4 |
- machine-generated
|
5 |
auto_acl:
|
6 |
- machine-generated
|
7 |
+
auto_full_no_split:
|
8 |
+
- machine-generated
|
9 |
+
auto_full_with_split:
|
10 |
+
- machine-generated
|
11 |
manual:
|
12 |
- crowdsourced
|
13 |
language_creators:
|
|
|
65 |
|
66 |
The authors first crowd-sourced a set of manual alignments between sentences in a subset of the Simple English Wikipedia and their corresponding versions in English Wikipedia (this corresponds to the `manual` config in this version of dataset), then trained a neural CRF system to predict these alignments.
|
67 |
|
68 |
+
The trained alignment prediction model was then applied to the other articles in Simple English Wikipedia with an English counterpart to create a larger corpus of aligned sentences (corresponding to the `auto`, `auto_acl`, `auto_full_no_split`, and `auto_full_with_split` configs here).
|
69 |
|
70 |
### Supported Tasks and Leaderboards
|
71 |
|
|
|
131 |
'simple_article_url': 'https://simple.wikipedia.org/wiki?curid=702227'}}
|
132 |
```
|
133 |
|
134 |
+
Finally, the `auto_acl`, the `auto_full_no_split`, and the `auto_full_with_split` configs were obtained by selecting the aligned pairs of sentences from `auto` to provide a ready-to-go aligned dataset to train a sequence-to-sequence system. While `auto_acl` corresponds to the filtered version of the data used to train the systems in the paper, `auto_full_no_split` and `auto_full_with_split` correspond to the unfiltered versions with and without sentence splits respectively. In the `auto_full_with_split` config, we join the sentences in the simple article mapped to the same sentence in the complex article to capture sentence splitting. Split sentences are seperated by a `<SEP>` token. In the `auto_full_no_split` config, we do not join the splits and treat them as seperate pairs. An instance is a single pair of sentences:
|
135 |
```
|
136 |
{'normal_sentence': 'In early work , Rutherford discovered the concept of radioactive half-life , the radioactive element radon , and differentiated and named alpha and beta radiation .\n',
|
137 |
'simple_sentence': 'Rutherford discovered the radioactive half-life , and the three parts of radiation which he named Alpha , Beta , and Gamma .\n'}
|
dataset_infos.json
CHANGED
@@ -1 +1 @@
|
|
1 |
-
{"manual": {"description": "WikiAuto provides a set of aligned sentences from English Wikipedia and Simple English Wikipedia\nas a resource to train sentence simplification systems. The authors first crowd-sourced a set of manual alignments\nbetween sentences in a subset of the Simple English Wikipedia and their corresponding versions in English Wikipedia\n(this corresponds to the `manual` config), then trained a neural CRF system to predict these alignments.\nThe trained model was then applied to the other articles in Simple English Wikipedia with an English counterpart to\ncreate a larger corpus of aligned sentences (corresponding to the `auto` and `auto_acl` configs here).\n", "citation": "@inproceedings{acl/JiangMLZX20,\n author = {Chao Jiang and\n Mounica Maddela and\n Wuwei Lan and\n Yang Zhong and\n Wei Xu},\n editor = {Dan Jurafsky and\n Joyce Chai and\n Natalie Schluter and\n Joel R. Tetreault},\n title = {Neural {CRF} Model for Sentence Alignment in Text Simplification},\n booktitle = {Proceedings of the 58th Annual Meeting of the Association for Computational\n Linguistics, {ACL} 2020, Online, July 5-10, 2020},\n pages = {7943--7960},\n publisher = {Association for Computational Linguistics},\n year = {2020},\n url = {https://www.aclweb.org/anthology/2020.acl-main.709/}\n}\n", "homepage": "https://github.com/chaojiang06/wiki-auto", "license": "CC-BY-SA 3.0", "features": {"alignment_label": {"num_classes": 2, "names": ["notAligned", "aligned"], "names_file": null, "id": null, "_type": "ClassLabel"}, "normal_sentence_id": {"dtype": "string", "id": null, "_type": "Value"}, "simple_sentence_id": {"dtype": "string", "id": null, "_type": "Value"}, "normal_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "simple_sentence": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "builder_name": "wiki_auto", "config_name": "manual", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 109343271, "num_examples": 373801, "dataset_name": "wiki_auto"}, "dev": {"name": "dev", "num_bytes": 20819779, "num_examples": 73249, "dataset_name": "wiki_auto"}, "test": {"name": "test", "num_bytes": 33379338, "num_examples": 118074, "dataset_name": "wiki_auto"}}, "download_checksums": {"https://github.com/chaojiang06/wiki-auto/raw/master/wiki-manual/train.tsv": {"num_bytes": 106346588, "checksum": "82fa388de3ded6d303b95fcd11ba70e0b6158d2df1cbf24913bb54503bd32e95"}, "https://github.com/chaojiang06/wiki-auto/raw/master/wiki-manual/dev.tsv": {"num_bytes": 20232621, "checksum": "c56a9d2a739f9da83f90c54e266e1d60dd036cb80c463f118cb55613232e2e41"}, "https://github.com/chaojiang06/wiki-auto/raw/master/wiki-manual/test.tsv": {"num_bytes": 32432523, "checksum": "ab8b818b0eeb7aa7712d244ee0ea16cfd915a896c40f02a34a808b597a5e68a0"}}, "download_size": 159011732, "post_processing_size": null, "dataset_size": 163542388, "size_in_bytes": 322554120}, "auto_acl": {"description": "WikiAuto provides a set of aligned sentences from English Wikipedia and Simple English Wikipedia\nas a resource to train sentence simplification systems. The authors first crowd-sourced a set of manual alignments\nbetween sentences in a subset of the Simple English Wikipedia and their corresponding versions in English Wikipedia\n(this corresponds to the `manual` config), then trained a neural CRF system to predict these alignments.\nThe trained model was then applied to the other articles in Simple English Wikipedia with an English counterpart to\ncreate a larger corpus of aligned sentences (corresponding to the `auto` and `auto_acl` configs here).\n", "citation": "@inproceedings{acl/JiangMLZX20,\n author = {Chao Jiang and\n Mounica Maddela and\n Wuwei Lan and\n Yang Zhong and\n Wei Xu},\n editor = {Dan Jurafsky and\n Joyce Chai and\n Natalie Schluter and\n Joel R. Tetreault},\n title = {Neural {CRF} Model for Sentence Alignment in Text Simplification},\n booktitle = {Proceedings of the 58th Annual Meeting of the Association for Computational\n Linguistics, {ACL} 2020, Online, July 5-10, 2020},\n pages = {7943--7960},\n publisher = {Association for Computational Linguistics},\n year = {2020},\n url = {https://www.aclweb.org/anthology/2020.acl-main.709/}\n}\n", "homepage": "https://github.com/chaojiang06/wiki-auto", "license": "CC-BY-SA 3.0", "features": {"normal_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "simple_sentence": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "builder_name": "wiki_auto", "config_name": "auto_acl", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"full": {"name": "full", "num_bytes": 121975414, "num_examples": 488332, "dataset_name": "wiki_auto"}}, "download_checksums": {"https://github.com/chaojiang06/wiki-auto/raw/master/wiki-auto/ACL2020/train.src": {"num_bytes": 70209062, "checksum": "02141edbb735be50c9942f5e0bced4528dc8d844753d46a1f3bdf0b6e550c0e6"}, "https://github.com/chaojiang06/wiki-auto/raw/master/wiki-auto/ACL2020/train.dst": {"num_bytes": 47859304, "checksum": "d9e2106722e2e29f34d5d9b697c236b38e920724727cefb71f42072dd9fd8807"}}, "download_size": 118068366, "post_processing_size": null, "dataset_size": 121975414, "size_in_bytes": 240043780}, "auto": {"description": "WikiAuto provides a set of aligned sentences from English Wikipedia and Simple English Wikipedia\nas a resource to train sentence simplification systems. The authors first crowd-sourced a set of manual alignments\nbetween sentences in a subset of the Simple English Wikipedia and their corresponding versions in English Wikipedia\n(this corresponds to the `manual` config), then trained a neural CRF system to predict these alignments.\nThe trained model was then applied to the other articles in Simple English Wikipedia with an English counterpart to\ncreate a larger corpus of aligned sentences (corresponding to the `auto` and `auto_acl` configs here).\n", "citation": "@inproceedings{acl/JiangMLZX20,\n author = {Chao Jiang and\n Mounica Maddela and\n Wuwei Lan and\n Yang Zhong and\n Wei Xu},\n editor = {Dan Jurafsky and\n Joyce Chai and\n Natalie Schluter and\n Joel R. Tetreault},\n title = {Neural {CRF} Model for Sentence Alignment in Text Simplification},\n booktitle = {Proceedings of the 58th Annual Meeting of the Association for Computational\n Linguistics, {ACL} 2020, Online, July 5-10, 2020},\n pages = {7943--7960},\n publisher = {Association for Computational Linguistics},\n year = {2020},\n url = {https://www.aclweb.org/anthology/2020.acl-main.709/}\n}\n", "homepage": "https://github.com/chaojiang06/wiki-auto", "license": "CC-BY-SA 3.0", "features": {"example_id": {"dtype": "string", "id": null, "_type": "Value"}, "normal": {"normal_article_id": {"dtype": "int32", "id": null, "_type": "Value"}, "normal_article_title": {"dtype": "string", "id": null, "_type": "Value"}, "normal_article_url": {"dtype": "string", "id": null, "_type": "Value"}, "normal_article_content": {"feature": {"normal_sentence_id": {"dtype": "string", "id": null, "_type": "Value"}, "normal_sentence": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "simple": {"simple_article_id": {"dtype": "int32", "id": null, "_type": "Value"}, "simple_article_title": {"dtype": "string", "id": null, "_type": "Value"}, "simple_article_url": {"dtype": "string", "id": null, "_type": "Value"}, "simple_article_content": {"feature": {"simple_sentence_id": {"dtype": "string", "id": null, "_type": "Value"}, "simple_sentence": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "paragraph_alignment": {"feature": {"normal_paragraph_id": {"dtype": "string", "id": null, "_type": "Value"}, "simple_paragraph_id": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}, "sentence_alignment": {"feature": {"normal_sentence_id": {"dtype": "string", "id": null, "_type": "Value"}, "simple_sentence_id": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "builder_name": "wiki_auto", "config_name": "auto", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"part_1": {"name": "part_1", "num_bytes": 1773240295, "num_examples": 125059, "dataset_name": "wiki_auto"}, "part_2": {"name": "part_2", "num_bytes": 80417651, "num_examples": 13036, "dataset_name": "wiki_auto"}}, "download_checksums": {"https://github.com/chaojiang06/wiki-auto/raw/master/wiki-auto/all_data/wiki-auto-part-1-data.json": {"num_bytes": 2067424750, "checksum": "136d8e113a773d3669228a57cae733fca079954daf0b3514505410c66d1a69b6"}, "https://github.com/chaojiang06/wiki-auto/raw/master/wiki-auto/all_data/wiki-auto-part-2-data.json": {"num_bytes": 93214171, "checksum": "94b33a11447c121a0ce7293de20fb969c36d8a62b31afc5873a4174ed17a1d4e"}}, "download_size": 2160638921, "post_processing_size": null, "dataset_size": 1853657946, "size_in_bytes": 4014296867}}
|
|
|
1 |
+
{"manual": {"description": "WikiAuto provides a set of aligned sentences from English Wikipedia and Simple English Wikipedia\nas a resource to train sentence simplification systems. The authors first crowd-sourced a set of manual alignments\nbetween sentences in a subset of the Simple English Wikipedia and their corresponding versions in English Wikipedia\n(this corresponds to the `manual` config), then trained a neural CRF system to predict these alignments.\nThe trained model was then applied to the other articles in Simple English Wikipedia with an English counterpart to\ncreate a larger corpus of aligned sentences (corresponding to the `auto`, `auto_acl`, `auto_full_no_split`, and `auto_full_with_split` configs here).\n", "citation": "@inproceedings{acl/JiangMLZX20,\n author = {Chao Jiang and\n Mounica Maddela and\n Wuwei Lan and\n Yang Zhong and\n Wei Xu},\n editor = {Dan Jurafsky and\n Joyce Chai and\n Natalie Schluter and\n Joel R. Tetreault},\n title = {Neural {CRF} Model for Sentence Alignment in Text Simplification},\n booktitle = {Proceedings of the 58th Annual Meeting of the Association for Computational\n Linguistics, {ACL} 2020, Online, July 5-10, 2020},\n pages = {7943--7960},\n publisher = {Association for Computational Linguistics},\n year = {2020},\n url = {https://www.aclweb.org/anthology/2020.acl-main.709/}\n}\n", "homepage": "https://github.com/chaojiang06/wiki-auto", "license": "CC-BY-SA 3.0", "features": {"alignment_label": {"num_classes": 2, "names": ["notAligned", "aligned"], "names_file": null, "id": null, "_type": "ClassLabel"}, "normal_sentence_id": {"dtype": "string", "id": null, "_type": "Value"}, "simple_sentence_id": {"dtype": "string", "id": null, "_type": "Value"}, "normal_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "simple_sentence": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "builder_name": "wiki_auto", "config_name": "manual", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 109343271, "num_examples": 373801, "dataset_name": "wiki_auto"}, "dev": {"name": "dev", "num_bytes": 20819779, "num_examples": 73249, "dataset_name": "wiki_auto"}, "test": {"name": "test", "num_bytes": 33379338, "num_examples": 118074, "dataset_name": "wiki_auto"}}, "download_checksums": {"https://github.com/chaojiang06/wiki-auto/raw/master/wiki-manual/train.tsv": {"num_bytes": 106346588, "checksum": "82fa388de3ded6d303b95fcd11ba70e0b6158d2df1cbf24913bb54503bd32e95"}, "https://github.com/chaojiang06/wiki-auto/raw/master/wiki-manual/dev.tsv": {"num_bytes": 20232621, "checksum": "c56a9d2a739f9da83f90c54e266e1d60dd036cb80c463f118cb55613232e2e41"}, "https://github.com/chaojiang06/wiki-auto/raw/master/wiki-manual/test.tsv": {"num_bytes": 32432523, "checksum": "ab8b818b0eeb7aa7712d244ee0ea16cfd915a896c40f02a34a808b597a5e68a0"}}, "download_size": 159011732, "post_processing_size": null, "dataset_size": 163542388, "size_in_bytes": 322554120}, "auto_acl": {"description": "WikiAuto provides a set of aligned sentences from English Wikipedia and Simple English Wikipedia\nas a resource to train sentence simplification systems. The authors first crowd-sourced a set of manual alignments\nbetween sentences in a subset of the Simple English Wikipedia and their corresponding versions in English Wikipedia\n(this corresponds to the `manual` config), then trained a neural CRF system to predict these alignments.\nThe trained model was then applied to the other articles in Simple English Wikipedia with an English counterpart to\ncreate a larger corpus of aligned sentences (corresponding to the `auto`, `auto_acl`, `auto_full_no_split`, and `auto_full_with_split` configs here).\n", "citation": "@inproceedings{acl/JiangMLZX20,\n author = {Chao Jiang and\n Mounica Maddela and\n Wuwei Lan and\n Yang Zhong and\n Wei Xu},\n editor = {Dan Jurafsky and\n Joyce Chai and\n Natalie Schluter and\n Joel R. Tetreault},\n title = {Neural {CRF} Model for Sentence Alignment in Text Simplification},\n booktitle = {Proceedings of the 58th Annual Meeting of the Association for Computational\n Linguistics, {ACL} 2020, Online, July 5-10, 2020},\n pages = {7943--7960},\n publisher = {Association for Computational Linguistics},\n year = {2020},\n url = {https://www.aclweb.org/anthology/2020.acl-main.709/}\n}\n", "homepage": "https://github.com/chaojiang06/wiki-auto", "license": "CC-BY-SA 3.0", "features": {"normal_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "simple_sentence": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "builder_name": "wiki_auto", "config_name": "auto_acl", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"full": {"name": "full", "num_bytes": 121975414, "num_examples": 488332, "dataset_name": "wiki_auto"}}, "download_checksums": {"https://github.com/chaojiang06/wiki-auto/raw/master/wiki-auto/ACL2020/train.src": {"num_bytes": 70209062, "checksum": "02141edbb735be50c9942f5e0bced4528dc8d844753d46a1f3bdf0b6e550c0e6"}, "https://github.com/chaojiang06/wiki-auto/raw/master/wiki-auto/ACL2020/train.dst": {"num_bytes": 47859304, "checksum": "d9e2106722e2e29f34d5d9b697c236b38e920724727cefb71f42072dd9fd8807"}}, "download_size": 118068366, "post_processing_size": null, "dataset_size": 121975414, "size_in_bytes": 240043780}, "auto": {"description": "WikiAuto provides a set of aligned sentences from English Wikipedia and Simple English Wikipedia\nas a resource to train sentence simplification systems. The authors first crowd-sourced a set of manual alignments\nbetween sentences in a subset of the Simple English Wikipedia and their corresponding versions in English Wikipedia\n(this corresponds to the `manual` config), then trained a neural CRF system to predict these alignments.\nThe trained model was then applied to the other articles in Simple English Wikipedia with an English counterpart to\ncreate a larger corpus of aligned sentences (corresponding to the `auto`, `auto_acl`, `auto_full_no_split`, and `auto_full_with_split` configs here).\n", "citation": "@inproceedings{acl/JiangMLZX20,\n author = {Chao Jiang and\n Mounica Maddela and\n Wuwei Lan and\n Yang Zhong and\n Wei Xu},\n editor = {Dan Jurafsky and\n Joyce Chai and\n Natalie Schluter and\n Joel R. Tetreault},\n title = {Neural {CRF} Model for Sentence Alignment in Text Simplification},\n booktitle = {Proceedings of the 58th Annual Meeting of the Association for Computational\n Linguistics, {ACL} 2020, Online, July 5-10, 2020},\n pages = {7943--7960},\n publisher = {Association for Computational Linguistics},\n year = {2020},\n url = {https://www.aclweb.org/anthology/2020.acl-main.709/}\n}\n", "homepage": "https://github.com/chaojiang06/wiki-auto", "license": "CC-BY-SA 3.0", "features": {"example_id": {"dtype": "string", "id": null, "_type": "Value"}, "normal": {"normal_article_id": {"dtype": "int32", "id": null, "_type": "Value"}, "normal_article_title": {"dtype": "string", "id": null, "_type": "Value"}, "normal_article_url": {"dtype": "string", "id": null, "_type": "Value"}, "normal_article_content": {"feature": {"normal_sentence_id": {"dtype": "string", "id": null, "_type": "Value"}, "normal_sentence": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "simple": {"simple_article_id": {"dtype": "int32", "id": null, "_type": "Value"}, "simple_article_title": {"dtype": "string", "id": null, "_type": "Value"}, "simple_article_url": {"dtype": "string", "id": null, "_type": "Value"}, "simple_article_content": {"feature": {"simple_sentence_id": {"dtype": "string", "id": null, "_type": "Value"}, "simple_sentence": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "paragraph_alignment": {"feature": {"normal_paragraph_id": {"dtype": "string", "id": null, "_type": "Value"}, "simple_paragraph_id": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}, "sentence_alignment": {"feature": {"normal_sentence_id": {"dtype": "string", "id": null, "_type": "Value"}, "simple_sentence_id": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "builder_name": "wiki_auto", "config_name": "auto", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"part_1": {"name": "part_1", "num_bytes": 1773240295, "num_examples": 125059, "dataset_name": "wiki_auto"}, "part_2": {"name": "part_2", "num_bytes": 80417651, "num_examples": 13036, "dataset_name": "wiki_auto"}}, "download_checksums": {"https://github.com/chaojiang06/wiki-auto/raw/master/wiki-auto/all_data/wiki-auto-part-1-data.json": {"num_bytes": 2067424750, "checksum": "136d8e113a773d3669228a57cae733fca079954daf0b3514505410c66d1a69b6"}, "https://github.com/chaojiang06/wiki-auto/raw/master/wiki-auto/all_data/wiki-auto-part-2-data.json": {"num_bytes": 93214171, "checksum": "94b33a11447c121a0ce7293de20fb969c36d8a62b31afc5873a4174ed17a1d4e"}}, "download_size": 2160638921, "post_processing_size": null, "dataset_size": 1853657946, "size_in_bytes": 4014296867}, "auto_full_no_split": {"description": "WikiAuto provides a set of aligned sentences from English Wikipedia and Simple English Wikipedia\nas a resource to train sentence simplification systems. The authors first crowd-sourced a set of manual alignments\nbetween sentences in a subset of the Simple English Wikipedia and their corresponding versions in English Wikipedia\n(this corresponds to the `manual` config), then trained a neural CRF system to predict these alignments.\nThe trained model was then applied to the other articles in Simple English Wikipedia with an English counterpart to\ncreate a larger corpus of aligned sentences (corresponding to the `auto`, `auto_acl`, `auto_full_no_split`, and `auto_full_with_split` configs here).\n", "citation": "@inproceedings{acl/JiangMLZX20,\n author = {Chao Jiang and\n Mounica Maddela and\n Wuwei Lan and\n Yang Zhong and\n Wei Xu},\n editor = {Dan Jurafsky and\n Joyce Chai and\n Natalie Schluter and\n Joel R. Tetreault},\n title = {Neural {CRF} Model for Sentence Alignment in Text Simplification},\n booktitle = {Proceedings of the 58th Annual Meeting of the Association for Computational\n Linguistics, {ACL} 2020, Online, July 5-10, 2020},\n pages = {7943--7960},\n publisher = {Association for Computational Linguistics},\n year = {2020},\n url = {https://www.aclweb.org/anthology/2020.acl-main.709/}\n}\n", "homepage": "https://github.com/chaojiang06/wiki-auto", "license": "CC-BY-SA 3.0", "features": {"normal_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "simple_sentence": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "builder_name": "wiki_auto", "config_name": "auto_full_no_split", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"full": {"name": "full", "num_bytes": 146310611, "num_examples": 591994, "dataset_name": "wiki_auto"}}, "download_checksums": {"https://github.com/chaojiang06/wiki-auto/raw/master/wiki-auto/GEM2021/full_no_split/train.src": {"num_bytes": 86028898, "checksum": "48d820fc1e7bd7121edde04a2aedf62bb9069b9caed01b77b7cb3852cc6b33d0"}, "https://github.com/chaojiang06/wiki-auto/raw/master/wiki-auto/GEM2021/full_no_split/train.dst": {"num_bytes": 55545281, "checksum": "37b373c5ee8c3f4b844c9896f07fb6bb55a20a3d4a8e1a6b06a651b87a653415"}}, "download_size": 141574179, "post_processing_size": null, "dataset_size": 146310611, "size_in_bytes": 287884790}, "auto_full_with_split": {"description": "WikiAuto provides a set of aligned sentences from English Wikipedia and Simple English Wikipedia\nas a resource to train sentence simplification systems. The authors first crowd-sourced a set of manual alignments\nbetween sentences in a subset of the Simple English Wikipedia and their corresponding versions in English Wikipedia\n(this corresponds to the `manual` config), then trained a neural CRF system to predict these alignments.\nThe trained model was then applied to the other articles in Simple English Wikipedia with an English counterpart to\ncreate a larger corpus of aligned sentences (corresponding to the `auto`, `auto_acl`, `auto_full_no_split`, and `auto_full_with_split` configs here).\n", "citation": "@inproceedings{acl/JiangMLZX20,\n author = {Chao Jiang and\n Mounica Maddela and\n Wuwei Lan and\n Yang Zhong and\n Wei Xu},\n editor = {Dan Jurafsky and\n Joyce Chai and\n Natalie Schluter and\n Joel R. Tetreault},\n title = {Neural {CRF} Model for Sentence Alignment in Text Simplification},\n booktitle = {Proceedings of the 58th Annual Meeting of the Association for Computational\n Linguistics, {ACL} 2020, Online, July 5-10, 2020},\n pages = {7943--7960},\n publisher = {Association for Computational Linguistics},\n year = {2020},\n url = {https://www.aclweb.org/anthology/2020.acl-main.709/}\n}\n", "homepage": "https://github.com/chaojiang06/wiki-auto", "license": "CC-BY-SA 3.0", "features": {"normal_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "simple_sentence": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "builder_name": "wiki_auto", "config_name": "auto_full_with_split", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"full": {"name": "full", "num_bytes": 124549115, "num_examples": 483801, "dataset_name": "wiki_auto"}}, "download_checksums": {"https://github.com/chaojiang06/wiki-auto/raw/master/wiki-auto/GEM2021/full_with_split/train.src": {"num_bytes": 67389841, "checksum": "f70dd6083bf679254dfd6c1e14960ad2b526bc35596d9c87b0c636c0285ce8ac"}, "https://github.com/chaojiang06/wiki-auto/raw/master/wiki-auto/GEM2021/full_with_split/train.dst": {"num_bytes": 53288474, "checksum": "74529cfc316c8a3ed7720f87e25d0758620e0e838a5747ec4a1bbdccef018f75"}}, "download_size": 120678315, "post_processing_size": null, "dataset_size": 124549115, "size_in_bytes": 245227430}}
|
dummy/auto_full_no_split/1.0.0/dummy_data.zip
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:1b0a237dd5adb45e4407eac11274b0da20fda345ad2f2acb0df8cfa725cc682e
|
3 |
+
size 983
|
dummy/auto_full_with_split/1.0.0/dummy_data.zip
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:c8600e06b2762e11ffcfaf9600edb585fb3a0fbfaadf7be1405281a408c5a344
|
3 |
+
size 1521
|
wiki_auto.py
CHANGED
@@ -50,7 +50,7 @@ as a resource to train sentence simplification systems. The authors first crowd-
|
|
50 |
between sentences in a subset of the Simple English Wikipedia and their corresponding versions in English Wikipedia
|
51 |
(this corresponds to the `manual` config), then trained a neural CRF system to predict these alignments.
|
52 |
The trained model was then applied to the other articles in Simple English Wikipedia with an English counterpart to
|
53 |
-
create a larger corpus of aligned sentences (corresponding to the `auto` and `
|
54 |
"""
|
55 |
|
56 |
# TODO: Add the licence for the dataset here if you can find it
|
@@ -69,6 +69,14 @@ _URLs = {
|
|
69 |
"normal": "https://github.com/chaojiang06/wiki-auto/raw/master/wiki-auto/ACL2020/train.src",
|
70 |
"simple": "https://github.com/chaojiang06/wiki-auto/raw/master/wiki-auto/ACL2020/train.dst",
|
71 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
72 |
"auto": {
|
73 |
"part_1": "https://github.com/chaojiang06/wiki-auto/raw/master/wiki-auto/all_data/wiki-auto-part-1-data.json",
|
74 |
"part_2": "https://github.com/chaojiang06/wiki-auto/raw/master/wiki-auto/all_data/wiki-auto-part-2-data.json",
|
@@ -89,7 +97,19 @@ class WikiAuto(datasets.GeneratorBasedBuilder):
|
|
89 |
description="A set of 10K Wikipedia sentence pairs aligned by crowd workers.",
|
90 |
),
|
91 |
datasets.BuilderConfig(
|
92 |
-
name="auto_acl",
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
93 |
),
|
94 |
datasets.BuilderConfig(
|
95 |
name="auto", version=VERSION, description="A large set of automatically aligned sentence pairs."
|
@@ -109,7 +129,11 @@ class WikiAuto(datasets.GeneratorBasedBuilder):
|
|
109 |
"simple_sentence": datasets.Value("string"),
|
110 |
}
|
111 |
)
|
112 |
-
elif
|
|
|
|
|
|
|
|
|
113 |
features = datasets.Features(
|
114 |
{
|
115 |
"normal_sentence": datasets.Value("string"),
|
@@ -201,7 +225,11 @@ class WikiAuto(datasets.GeneratorBasedBuilder):
|
|
201 |
values = line.strip().split("\t")
|
202 |
assert len(values) == 5, f"Not enough fields in ---- {line} --- {values}"
|
203 |
yield id_, dict([(k, val) for k, val in zip(keys, values)])
|
204 |
-
elif
|
|
|
|
|
|
|
|
|
205 |
with open(filepaths["normal"], encoding="utf-8") as fi:
|
206 |
with open(filepaths["simple"], encoding="utf-8") as fo:
|
207 |
for id_, (norm_se, simp_se) in enumerate(zip(fi, fo)):
|
|
|
50 |
between sentences in a subset of the Simple English Wikipedia and their corresponding versions in English Wikipedia
|
51 |
(this corresponds to the `manual` config), then trained a neural CRF system to predict these alignments.
|
52 |
The trained model was then applied to the other articles in Simple English Wikipedia with an English counterpart to
|
53 |
+
create a larger corpus of aligned sentences (corresponding to the `auto`, `auto_acl`, `auto_full_no_split`, and `auto_full_with_split` configs here).
|
54 |
"""
|
55 |
|
56 |
# TODO: Add the licence for the dataset here if you can find it
|
|
|
69 |
"normal": "https://github.com/chaojiang06/wiki-auto/raw/master/wiki-auto/ACL2020/train.src",
|
70 |
"simple": "https://github.com/chaojiang06/wiki-auto/raw/master/wiki-auto/ACL2020/train.dst",
|
71 |
},
|
72 |
+
"auto_full_no_split": {
|
73 |
+
"normal": "https://github.com/chaojiang06/wiki-auto/raw/master/wiki-auto/GEM2021/full_no_split/train.src",
|
74 |
+
"simple": "https://github.com/chaojiang06/wiki-auto/raw/master/wiki-auto/GEM2021/full_no_split/train.dst",
|
75 |
+
},
|
76 |
+
"auto_full_with_split": {
|
77 |
+
"normal": "https://github.com/chaojiang06/wiki-auto/raw/master/wiki-auto/GEM2021/full_with_split/train.src",
|
78 |
+
"simple": "https://github.com/chaojiang06/wiki-auto/raw/master/wiki-auto/GEM2021/full_with_split/train.dst",
|
79 |
+
},
|
80 |
"auto": {
|
81 |
"part_1": "https://github.com/chaojiang06/wiki-auto/raw/master/wiki-auto/all_data/wiki-auto-part-1-data.json",
|
82 |
"part_2": "https://github.com/chaojiang06/wiki-auto/raw/master/wiki-auto/all_data/wiki-auto-part-2-data.json",
|
|
|
97 |
description="A set of 10K Wikipedia sentence pairs aligned by crowd workers.",
|
98 |
),
|
99 |
datasets.BuilderConfig(
|
100 |
+
name="auto_acl",
|
101 |
+
version=VERSION,
|
102 |
+
description="Automatically aligned and filtered sentence pairs used to train the ACL2020 system.",
|
103 |
+
),
|
104 |
+
datasets.BuilderConfig(
|
105 |
+
name="auto_full_no_split",
|
106 |
+
version=VERSION,
|
107 |
+
description="All automatically aligned sentence pairs without sentence splitting.",
|
108 |
+
),
|
109 |
+
datasets.BuilderConfig(
|
110 |
+
name="auto_full_with_split",
|
111 |
+
version=VERSION,
|
112 |
+
description="All automatically aligned sentence pairs with sentence splitting.",
|
113 |
),
|
114 |
datasets.BuilderConfig(
|
115 |
name="auto", version=VERSION, description="A large set of automatically aligned sentence pairs."
|
|
|
129 |
"simple_sentence": datasets.Value("string"),
|
130 |
}
|
131 |
)
|
132 |
+
elif (
|
133 |
+
self.config.name == "auto_acl"
|
134 |
+
or self.config.name == "auto_full_no_split"
|
135 |
+
or self.config.name == "auto_full_with_split"
|
136 |
+
):
|
137 |
features = datasets.Features(
|
138 |
{
|
139 |
"normal_sentence": datasets.Value("string"),
|
|
|
225 |
values = line.strip().split("\t")
|
226 |
assert len(values) == 5, f"Not enough fields in ---- {line} --- {values}"
|
227 |
yield id_, dict([(k, val) for k, val in zip(keys, values)])
|
228 |
+
elif (
|
229 |
+
self.config.name == "auto_acl"
|
230 |
+
or self.config.name == "auto_full_no_split"
|
231 |
+
or self.config.name == "auto_full_with_split"
|
232 |
+
):
|
233 |
with open(filepaths["normal"], encoding="utf-8") as fi:
|
234 |
with open(filepaths["simple"], encoding="utf-8") as fo:
|
235 |
for id_, (norm_se, simp_se) in enumerate(zip(fi, fo)):
|