Datasets:
lmqg
/

Modalities:
Text
Languages:
Korean
ArXiv:
Libraries:
Datasets
License:
qg_koquad / README.md
albertvillanova's picture
Fix task arrays
9bff4f4
|
raw
history blame
6.53 kB
metadata
license: cc-by-4.0
pretty_name: KorQuAD for question generation
language: ko
multilinguality: monolingual
size_categories: 10K<n<100K
source_datasets: squad_es
task_categories:
  - text-generation
task_ids:
  - language-modeling
tags:
  - question-generation

Dataset Card for "lmqg/qg_korquad"

Dataset Description

Dataset Summary

This is a subset of QG-Bench, a unified question generation benchmark proposed in "Generative Language Models for Paragraph-Level Question Generation: A Unified Benchmark and Evaluation, EMNLP 2022 main conference". This is a modified version of KorQuAD for question generation (QG) task. Since the original dataset only contains training/validation set, we manually sample test set from training set, which has no overlap in terms of the paragraph with the training set.

Supported Tasks and Leaderboards

  • question-generation: The dataset is assumed to be used to train a model for question generation. Success on this task is typically measured by achieving a high BLEU4/METEOR/ROUGE-L/BERTScore/MoverScore (see our paper for more in detail).

Languages

Korean (ko)

Dataset Structure

An example of 'train' looks as follows.

{
  "question": "ν•¨μˆ˜ν•΄μ„ν•™μ΄ μ£Όλͺ©ν•˜λŠ” νƒκ΅¬λŠ”?",
  "paragraph": "변화에 λŒ€ν•œ 이해와 λ¬˜μ‚¬λŠ” μžμ—°κ³Όν•™μ— μžˆμ–΄μ„œ 일반적인 주제이며, 미적뢄학은 λ³€ν™”λ₯Ό νƒκ΅¬ν•˜λŠ” κ°•λ ₯ν•œ λ„κ΅¬λ‘œμ„œ λ°œμ „λ˜μ—ˆλ‹€. ν•¨μˆ˜λŠ” λ³€ν™”ν•˜λŠ” 양을 λ¬˜μ‚¬ν•¨μ— μžˆμ–΄μ„œ 쀑좔적인 κ°œλ…μœΌλ‘œμ¨ λ– μ˜€λ₯΄κ²Œ λœλ‹€. μ‹€μˆ˜μ™€ μ‹€λ³€μˆ˜λ‘œ κ΅¬μ„±λœ ν•¨μˆ˜μ˜ μ—„λ°€ν•œ 탐ꡬ가 μ‹€ν•΄μ„ν•™μ΄λΌλŠ” λΆ„μ•Όλ‘œ μ•Œλ €μ§€κ²Œ λ˜μ—ˆκ³ , λ³΅μ†Œμˆ˜μ— λŒ€ν•œ 이와 같은 νƒκ΅¬λΆ„μ•ΌλŠ” λ³΅μ†Œν•΄μ„ν•™μ΄λΌκ³  ν•œλ‹€. ν•¨μˆ˜ν•΄μ„ν•™μ€ ν•¨μˆ˜μ˜ 곡간(특히 λ¬΄ν•œμ°¨μ›)의 탐ꡬ에 μ£Όλͺ©ν•œλ‹€. ν•¨μˆ˜ν•΄μ„ν•™μ˜ λ§Žμ€ μ‘μš©λΆ„μ•Ό 쀑 ν•˜λ‚˜κ°€ μ–‘μžμ—­ν•™μ΄λ‹€. λ§Žμ€ λ¬Έμ œλ“€μ΄ μžμ—°μŠ€λŸ½κ²Œ μ–‘κ³Ό κ·Έ μ–‘μ˜ λ³€ν™”μœ¨μ˜ κ΄€κ³„λ‘œ κ·€μ°©λ˜κ³ , μ΄λŸ¬ν•œ λ¬Έμ œλ“€μ΄ λ―ΈλΆ„λ°©μ •μ‹μœΌλ‘œ 닀루어진닀. μžμ—°μ˜ λ§Žμ€ ν˜„μƒλ“€μ΄ λ™μ—­ν•™κ³„λ‘œ 기술될 수 μžˆλ‹€. 혼돈 이둠은 μ΄λŸ¬ν•œ 예츑 λΆˆκ°€λŠ₯ν•œ ν˜„μƒμ„ νƒκ΅¬ν•˜λŠ” 데 μƒλ‹Ήν•œ κΈ°μ—¬λ₯Ό ν•œλ‹€.",
  "answer": "ν•¨μˆ˜μ˜ 곡간(특히 λ¬΄ν•œμ°¨μ›)의 탐ꡬ",
  "sentence": "ν•¨μˆ˜ν•΄μ„ν•™μ€ ν•¨μˆ˜μ˜ 곡간(특히 λ¬΄ν•œμ°¨μ›)의 탐ꡬ 에 μ£Όλͺ©ν•œλ‹€.",
  "paragraph_sentence": '변화에 λŒ€ν•œ 이해와 λ¬˜μ‚¬λŠ” μžμ—°κ³Όν•™μ— μžˆμ–΄μ„œ 일반적인 주제이며, 미적뢄학은 λ³€ν™”λ₯Ό νƒκ΅¬ν•˜λŠ” κ°•λ ₯ν•œ λ„κ΅¬λ‘œμ„œ λ°œμ „λ˜μ—ˆλ‹€. ν•¨μˆ˜λŠ” λ³€ν™”ν•˜λŠ” 양을 λ¬˜μ‚¬ν•¨μ— μžˆμ–΄μ„œ 쀑좔적인 κ°œλ…μœΌλ‘œμ¨ λ– μ˜€λ₯΄κ²Œ λœλ‹€. μ‹€μˆ˜μ™€ μ‹€λ³€μˆ˜λ‘œ κ΅¬μ„±λœ ν•¨μˆ˜μ˜ μ—„λ°€ν•œ 탐ꡬ가 μ‹€ν•΄μ„ν•™μ΄λΌλŠ” λΆ„μ•Όλ‘œ μ•Œλ €μ§€κ²Œ λ˜μ—ˆκ³ , λ³΅μ†Œμˆ˜μ— λŒ€ν•œ 이와 같은 탐ꡬ λΆ„μ•ΌλŠ” λ³΅μ†Œν•΄μ„ν•™μ΄λΌκ³  ν•œλ‹€. <hl> ν•¨μˆ˜ν•΄μ„ν•™μ€ ν•¨μˆ˜μ˜ 곡간(특히 λ¬΄ν•œμ°¨μ›)의 탐ꡬ 에 μ£Όλͺ©ν•œλ‹€. <hl> ν•¨μˆ˜ν•΄μ„ν•™μ˜ λ§Žμ€ μ‘μš©λΆ„μ•Ό 쀑 ν•˜λ‚˜κ°€ μ–‘μžμ—­ν•™μ΄λ‹€. λ§Žμ€ λ¬Έμ œλ“€μ΄ μžμ—°μŠ€λŸ½κ²Œ μ–‘κ³Ό κ·Έ μ–‘μ˜ λ³€ν™”μœ¨μ˜ κ΄€κ³„λ‘œ κ·€μ°©λ˜κ³ , μ΄λŸ¬ν•œ λ¬Έμ œλ“€μ΄ λ―ΈλΆ„λ°©μ •μ‹μœΌλ‘œ 닀루어진닀. μžμ—°μ˜ λ§Žμ€ ν˜„μƒλ“€μ΄ λ™μ—­ν•™κ³„λ‘œ 기술될 수 μžˆλ‹€. 혼돈 이둠은 μ΄λŸ¬ν•œ 예츑 λΆˆκ°€λŠ₯ν•œ ν˜„μƒμ„ νƒκ΅¬ν•˜λŠ” 데 μƒλ‹Ήν•œ κΈ°μ—¬λ₯Ό ν•œλ‹€.',
  "paragraph_answer": '변화에 λŒ€ν•œ 이해와 λ¬˜μ‚¬λŠ” μžμ—°κ³Όν•™μ— μžˆμ–΄μ„œ 일반적인 주제이며, 미적뢄학은 λ³€ν™”λ₯Ό νƒκ΅¬ν•˜λŠ” κ°•λ ₯ν•œ λ„κ΅¬λ‘œμ„œ λ°œμ „λ˜μ—ˆλ‹€. ν•¨μˆ˜λŠ” λ³€ν™”ν•˜λŠ” 양을 λ¬˜μ‚¬ν•¨μ— μžˆμ–΄μ„œ 쀑좔적인 κ°œλ…μœΌλ‘œμ¨ λ– μ˜€λ₯΄κ²Œ λœλ‹€. μ‹€μˆ˜μ™€ μ‹€λ³€μˆ˜λ‘œ κ΅¬μ„±λœ ν•¨μˆ˜μ˜ μ—„λ°€ν•œ 탐ꡬ가 μ‹€ν•΄μ„ν•™μ΄λΌλŠ” λΆ„μ•Όλ‘œ μ•Œλ €μ§€κ²Œ λ˜μ—ˆκ³ , λ³΅μ†Œμˆ˜μ— λŒ€ν•œ 이와 같은 탐ꡬ λΆ„μ•ΌλŠ” λ³΅μ†Œν•΄μ„ν•™μ΄λΌκ³  ν•œλ‹€. ν•¨μˆ˜ν•΄μ„ν•™μ€ <hl> ν•¨μˆ˜μ˜ 곡간(특히 λ¬΄ν•œμ°¨μ›)의 탐ꡬ <hl>에 μ£Όλͺ©ν•œλ‹€. ν•¨μˆ˜ν•΄μ„ν•™μ˜ λ§Žμ€ μ‘μš©λΆ„μ•Ό 쀑 ν•˜λ‚˜κ°€ μ–‘μžμ—­ν•™μ΄λ‹€. λ§Žμ€ λ¬Έμ œλ“€μ΄ μžμ—°μŠ€λŸ½κ²Œ μ–‘κ³Ό κ·Έ μ–‘μ˜ λ³€ν™”μœ¨μ˜ κ΄€κ³„λ‘œ κ·€μ°©λ˜κ³ , μ΄λŸ¬ν•œ λ¬Έμ œλ“€μ΄ λ―ΈλΆ„λ°©μ •μ‹μœΌλ‘œ 닀루어진닀. μžμ—°μ˜ λ§Žμ€ ν˜„μƒλ“€μ΄ λ™μ—­ν•™κ³„λ‘œ 기술될 수 μžˆλ‹€. 혼돈 이둠은 μ΄λŸ¬ν•œ 예츑 λΆˆκ°€λŠ₯ν•œ ν˜„μƒμ„ νƒκ΅¬ν•˜λŠ” 데 μƒλ‹Ήν•œ κΈ°μ—¬λ₯Ό ν•œλ‹€.',
  "sentence_answer": "ν•¨μˆ˜ν•΄μ„ν•™μ€ <hl> ν•¨μˆ˜μ˜ 곡간(특히 λ¬΄ν•œμ°¨μ›)의 탐ꡬ <hl> 에 μ£Όλͺ©ν•œλ‹€."
}

The data fields are the same among all splits.

  • question: a string feature.
  • paragraph: a string feature.
  • answer: a string feature.
  • sentence: a string feature.
  • paragraph_answer: a string feature, which is same as the paragraph but the answer is highlighted by a special token <hl>.
  • paragraph_sentence: a string feature, which is same as the paragraph but a sentence containing the answer is highlighted by a special token <hl>.
  • sentence_answer: a string feature, which is same as the sentence but the answer is highlighted by a special token <hl>.

Each of paragraph_answer, paragraph_sentence, and sentence_answer feature is assumed to be used to train a question generation model, but with different information. The paragraph_answer and sentence_answer features are for answer-aware question generation and paragraph_sentence feature is for sentence-aware question generation.

Data Splits

train validation test
54556 5766 5766

Citation Information

@inproceedings{ushio-etal-2022-generative,
    title = "{G}enerative {L}anguage {M}odels for {P}aragraph-{L}evel {Q}uestion {G}eneration: {A} {U}nified {B}enchmark and {E}valuation",
    author = "Ushio, Asahi  and
        Alva-Manchego, Fernando  and
        Camacho-Collados, Jose",
    booktitle = "Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing",
    month = dec,
    year = "2022",
    address = "Abu Dhabi, U.A.E.",
    publisher = "Association for Computational Linguistics",
}