Google I/O is a wrap! Catch up on TensorFlow sessions View sessions

wiki_auto

  • Description:

WikiAuto provides a set of aligned sentences from English Wikipedia and Simple English Wikipedia as a resource to train sentence simplification systems. 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), then trained a neural CRF system to predict these alignments. The trained 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).

@inproceedings{acl/JiangMLZX20,
  author    = {Chao Jiang and
               Mounica Maddela and
               Wuwei Lan and
               Yang Zhong and
               Wei Xu},
  editor    = {Dan Jurafsky and
               Joyce Chai and
               Natalie Schluter and
               Joel R. Tetreault},
  title     = {Neural {CRF} Model for Sentence Alignment in Text Simplification},
  booktitle = {Proceedings of the 58th Annual Meeting of the Association for Computational
               Linguistics, {ACL} 2020, Online, July 5-10, 2020},
  pages     = {7943--7960},
  publisher = {Association for Computational Linguistics},
  year      = {2020},
  url       = {https://www.aclweb.org/anthology/2020.acl-main.709/}
}

wiki_auto/manual (default config)

  • Config description: A set of 10K Wikipedia sentence pairs aligned by crowd workers.

  • Download size: 53.47 MiB

  • Dataset size: 76.87 MiB

  • Auto-cached (documentation): Yes

  • Splits:

Split Examples
'dev' 73,249
'test' 118,074
  • Feature structure:
FeaturesDict({
    'GLEU-score': tf.float64,
    'alignment_label': ClassLabel(shape=(), dtype=tf.int64, num_classes=3),
    'normal_sentence': Text(shape=(), dtype=tf.string),
    'normal_sentence_id': Text(shape=(), dtype=tf.string),
    'simple_sentence': Text(shape=(), dtype=tf.string),
    'simple_sentence_id': Text(shape=(), dtype=tf.string),
})
  • Feature documentation:
Feature Class Shape Dtype Description
FeaturesDict
GLEU-score Tensor tf.float64
alignment_label ClassLabel tf.int64
normal_sentence Text tf.string
normal_sentence_id Text tf.string
simple_sentence Text tf.string
simple_sentence_id Text tf.string

wiki_auto/auto_acl

  • Config description: Sentence pairs aligned to train the ACL2020 system.

  • Download size: 112.60 MiB

  • Dataset size: 138.83 MiB

  • Auto-cached (documentation): Only when shuffle_files=False (full)

  • Splits:

Split Examples
'full' 488,332
  • Feature structure:
FeaturesDict({
    'normal_sentence': Text(shape=(), dtype=tf.string),
    'simple_sentence': Text(shape=(), dtype=tf.string),
})
  • Feature documentation:
Feature Class Shape Dtype Description
FeaturesDict
normal_sentence Text tf.string
simple_sentence Text tf.string

wiki_auto/auto_full_no_split

  • Config description: All automatically aligned sentence pairs without sentence splitting.

  • Download size: 135.02 MiB

  • Dataset size: 166.78 MiB

  • Auto-cached (documentation): Only when shuffle_files=False (full)

  • Splits:

Split Examples
'full' 591,994
  • Feature structure:
FeaturesDict({
    'normal_sentence': Text(shape=(), dtype=tf.string),
    'simple_sentence': Text(shape=(), dtype=tf.string),
})
  • Feature documentation:
Feature Class Shape Dtype Description
FeaturesDict
normal_sentence Text tf.string
simple_sentence Text tf.string

wiki_auto/auto_full_with_split

  • Config description: All automatically aligned sentence pairs with sentence splitting.

  • Download size: 115.09 MiB

  • Dataset size: 141.20 MiB

  • Auto-cached (documentation): Only when shuffle_files=False (full)

  • Splits:

Split Examples
'full' 483,801
  • Feature structure:
FeaturesDict({
    'normal_sentence': Text(shape=(), dtype=tf.string),
    'simple_sentence': Text(shape=(), dtype=tf.string),
})
  • Feature documentation:
Feature Class Shape Dtype Description
FeaturesDict
normal_sentence Text tf.string
simple_sentence Text tf.string

wiki_auto/auto

  • Config description: A large set of automatically aligned sentence pairs.

  • Download size: 2.01 GiB

  • Dataset size: 1.76 GiB

  • Auto-cached (documentation): No

  • Splits:

Split Examples
'part_1' 125,059
'part_2' 13,036
  • Feature structure:
FeaturesDict({
    'example_id': Text(shape=(), dtype=tf.string),
    'normal': FeaturesDict({
        'normal_article_content': Sequence({
            'normal_sentence': Text(shape=(), dtype=tf.string),
            'normal_sentence_id': Text(shape=(), dtype=tf.string),
        }),
        'normal_article_id': tf.int32,
        'normal_article_title': Text(shape=(), dtype=tf.string),
        'normal_article_url': Text(shape=(), dtype=tf.string),
    }),
    'paragraph_alignment': Sequence({
        'normal_paragraph_id': Text(shape=(), dtype=tf.string),
        'simple_paragraph_id': Text(shape=(), dtype=tf.string),
    }),
    'sentence_alignment': Sequence({
        'normal_sentence_id': Text(shape=(), dtype=tf.string),
        'simple_sentence_id': Text(shape=(), dtype=tf.string),
    }),
    'simple': FeaturesDict({
        'simple_article_content': Sequence({
            'simple_sentence': Text(shape=(), dtype=tf.string),
            'simple_sentence_id': Text(shape=(), dtype=tf.string),
        }),
        'simple_article_id': tf.int32,
        'simple_article_title': Text(shape=(), dtype=tf.string),
        'simple_article_url': Text(shape=(), dtype=tf.string),
    }),
})
  • Feature documentation:
Feature Class Shape Dtype Description
FeaturesDict
example_id Text tf.string
normal FeaturesDict
normal/normal_article_content Sequence
normal/normal_article_content/normal_sentence Text tf.string
normal/normal_article_content/normal_sentence_id Text tf.string
normal/normal_article_id Tensor tf.int32
normal/normal_article_title Text tf.string
normal/normal_article_url Text tf.string
paragraph_alignment Sequence
paragraph_alignment/normal_paragraph_id Text tf.string
paragraph_alignment/simple_paragraph_id Text tf.string
sentence_alignment Sequence
sentence_alignment/normal_sentence_id Text tf.string
sentence_alignment/simple_sentence_id Text tf.string
simple FeaturesDict
simple/simple_article_content Sequence
simple/simple_article_content/simple_sentence Text tf.string
simple/simple_article_content/simple_sentence_id Text tf.string
simple/simple_article_id Tensor tf.int32
simple/simple_article_title Text tf.string
simple/simple_article_url Text tf.string