Hugging Face's logo

Datasets:
matteogabburo
/
mWikiQA

Tasks:
Question Answering
Modalities:
Tabular
Text
Formats:
json
Languages:
English
French
German
Size:
100K - 1M
ArXiv:
Libraries:
Datasets
pandas
License:
Dataset card Viewer Files Community
1
mWikiQA / README.md
matteogabburo's picture
matteogabburo
Update README.md
3960a85 verified
preview code
|
raw
history blame contribute delete
9.87 kB
metadata
license: other
license_name: other
license_link: https://huggingface.co/datasets/microsoft/wiki_qa#licensing-information
task_categories:
  - question-answering
language:
  - en
  - fr
  - de
  - it
  - es
  - pt
pretty_name: mWikiQA
size_categories:
  - 100K<n<1M
configs:
  - config_name: default
    data_files:
      - split: train_en
        path: eng_train.jsonl
      - split: train_de
        path: deu_train.jsonl
      - split: train_fr
        path: fra_train.jsonl
      - split: train_it
        path: ita_train.jsonl
      - split: train_po
        path: por_train.jsonl
      - split: train_sp
        path: spa_train.jsonl
      - split: validation_en
        path: eng_dev.jsonl
      - split: validation_de
        path: deu_dev.jsonl
      - split: validation_fr
        path: fra_dev.jsonl
      - split: validation_it
        path: ita_dev.jsonl
      - split: validation_po
        path: por_dev.jsonl
      - split: validation_sp
        path: spa_dev.jsonl
      - split: test_en
        path: eng_test.jsonl
      - split: test_de
        path: deu_test.jsonl
      - split: test_fr
        path: fra_test.jsonl
      - split: test_it
        path: ita_test.jsonl
      - split: test_po
        path: por_test.jsonl
      - split: test_sp
        path: spa_test.jsonl
  - config_name: clean
    data_files:
      - split: train_en
        path: eng_train.jsonl
      - split: train_de
        path: deu_train.jsonl
      - split: train_fr
        path: fra_train.jsonl
      - split: train_it
        path: ita_train.jsonl
      - split: train_po
        path: por_train.jsonl
      - split: train_sp
        path: spa_train.jsonl
      - split: validation_clean_en
        path: eng_dev_clean.jsonl
      - split: validation_clean_de
        path: deu_dev_clean.jsonl
      - split: validation_clean_fr
        path: fra_dev_clean.jsonl
      - split: validation_clean_it
        path: ita_dev_clean.jsonl
      - split: validation_clean_po
        path: por_dev_clean.jsonl
      - split: validation_clean_sp
        path: spa_dev_clean.jsonl
      - split: test_clean_en
        path: eng_test_clean.jsonl
      - split: test_clean_de
        path: deu_test_clean.jsonl
      - split: test_clean_fr
        path: fra_test_clean.jsonl
      - split: test_clean_it
        path: ita_test_clean.jsonl
      - split: test_clean_po
        path: por_test_clean.jsonl
      - split: test_clean_sp
        path: spa_test_clean.jsonl
  - config_name: noneg
    data_files:
      - split: train_en
        path: eng_train.jsonl
      - split: train_de
        path: deu_train.jsonl
      - split: train_fr
        path: fra_train.jsonl
      - split: train_it
        path: ita_train.jsonl
      - split: train_po
        path: por_train.jsonl
      - split: train_sp
        path: spa_train.jsonl
      - split: validation_noneg_en
        path: eng_dev_no_allneg.jsonl
      - split: validation_noneg_de
        path: deu_dev_no_allneg.jsonl
      - split: validation_noneg_fr
        path: fra_dev_no_allneg.jsonl
      - split: validation_noneg_it
        path: ita_dev_no_allneg.jsonl
      - split: validation_noneg_po
        path: por_dev_no_allneg.jsonl
      - split: validation_noneg_sp
        path: spa_dev_no_allneg.jsonl
      - split: test_noneg_en
        path: eng_test_no_allneg.jsonl
      - split: test_noneg_de
        path: deu_test_no_allneg.jsonl
      - split: test_noneg_fr
        path: fra_test_no_allneg.jsonl
      - split: test_noneg_it
        path: ita_test_no_allneg.jsonl
      - split: test_noneg_po
        path: por_test_no_allneg.jsonl
      - split: test_noneg_sp
        path: spa_test_no_allneg.jsonl
  - config_name: en
    data_files:
      - split: train
        path: eng_train.jsonl
      - split: validation
        path: eng_dev.jsonl
      - split: test
        path: eng_test.jsonl
  - config_name: de
    data_files:
      - split: train
        path: deu_train.jsonl
      - split: validation
        path: deu_dev.jsonl
      - split: test
        path: deu_test.jsonl
  - config_name: fr
    data_files:
      - split: train
        path: fra_train.jsonl
      - split: validation
        path: fra_dev.jsonl
      - split: test
        path: fra_test.jsonl
  - config_name: it
    data_files:
      - split: train
        path: ita_train.jsonl
      - split: validation
        path: ita_dev.jsonl
      - split: test
        path: ita_test.jsonl
  - config_name: po
    data_files:
      - split: train
        path: por_train.jsonl
      - split: validation
        path: por_dev.jsonl
      - split: test
        path: por_test.jsonl
  - config_name: sp
    data_files:
      - split: train
        path: spa_train.jsonl
      - split: validation
        path: spa_dev.jsonl
      - split: test
        path: spa_test.jsonl
  - config_name: en_noneg
    data_files:
      - split: train
        path: eng_train.jsonl
      - split: validation
        path: eng_dev_no_allneg.jsonl
      - split: test
        path: eng_test_no_allneg.jsonl
  - config_name: de_noneg
    data_files:
      - split: train
        path: deu_train.jsonl
      - split: validation
        path: deu_dev_no_allneg.jsonl
      - split: test
        path: deu_test_no_allneg.jsonl
  - config_name: fr_noneg
    data_files:
      - split: train
        path: fra_train.jsonl
      - split: validation
        path: fra_dev_no_allneg.jsonl
      - split: test
        path: fra_test_no_allneg.jsonl
  - config_name: it_noneg
    data_files:
      - split: train
        path: ita_train.jsonl
      - split: validation
        path: ita_dev_no_allneg.jsonl
      - split: test
        path: ita_test_no_allneg.jsonl
  - config_name: po_noneg
    data_files:
      - split: train
        path: por_train.jsonl
      - split: validation
        path: por_dev_no_allneg.jsonl
      - split: test
        path: por_test_no_allneg.jsonl
  - config_name: sp_noneg
    data_files:
      - split: train
        path: spa_train.jsonl
      - split: validation
        path: spa_dev_no_allneg.jsonl
      - split: test
        path: spa_test_no_allneg.jsonl
  - config_name: en_clean
    data_files:
      - split: train
        path: eng_train.jsonl
      - split: validation
        path: eng_dev_clean.jsonl
      - split: test
        path: eng_test_clean.jsonl
  - config_name: de_clean
    data_files:
      - split: train
        path: deu_train.jsonl
      - split: validation
        path: deu_dev_clean.jsonl
      - split: test
        path: deu_test_clean.jsonl
  - config_name: fr_clean
    data_files:
      - split: train
        path: fra_train.jsonl
      - split: validation
        path: fra_dev_clean.jsonl
      - split: test
        path: fra_test_clean.jsonl
  - config_name: it_clean
    data_files:
      - split: train
        path: ita_train.jsonl
      - split: validation
        path: ita_dev_clean.jsonl
      - split: test
        path: ita_test_clean.jsonl
  - config_name: po_clean
    data_files:
      - split: train
        path: por_train.jsonl
      - split: validation
        path: por_dev_clean.jsonl
      - split: test
        path: por_test_clean.jsonl
  - config_name: sp_clean
    data_files:
      - split: train
        path: spa_train.jsonl
      - split: validation
        path: spa_dev_clean.jsonl
      - split: test
        path: spa_test_clean.jsonl

Dataset Description

mWikiQA is a translated version of WikiQA. It contains 3,047 questions sampled from Bing query logs. The candidate answer sentences are extracted from Wikipedia and then manually labeled to assess whether they are correct answers.

The dataset has been translated into five European languages: French, German, Italian, Portuguese, and Spanish, as described in this paper: Datasets for Multilingual Answer Sentence Selection.

Splits:

For each language (English, French, German, Italian, Portuguese, and Spanish), we provide:

  • train split
  • validation split
  • test split

In addition, the validation and the test splits are available also in the following preprocessed versions:

  • noneg: without questions with only negative answer candidates
  • clean: without questions with only negative and only positive answer candidates

How to load them:

To use these splits, you can use the following snippet of code replacing [LANG] with a language identifier (en, fr, de, it, po, sp), and [VERSION] with the version identifier (noneg, clean)

from datasets import load_dataset

# if you want the whole corpora
corpora = load_dataset("matteogabburo/mWikiQA")

# if you want the clean test and test sets
corpora = load_dataset("matteogabburo/mWikiQA", "clean")

# if you want the "no all negatives" validation and test sets
corpora = load_dataset("matteogabburo/mWikiQA", "noneg")

"""
if you want the default splits of a specific language, replace [LANG] with an identifier in: en, fr, de, it, po, sp
dataset = load_dataset("matteogabburo/mWikiQA", "[LANG]")
"""
# example:
italian_dataset = load_dataset("matteogabburo/mWikiQA", "it")


"""
if you want the processed splits ("clean" and "no all negatives" sets), replace [LANG] with a language identifier and [VERSION] with "noneg" or "clean"
dataset = load_dataset("matteogabburo/mWikiQA", "[LANG]_[VERSION]")
"""
# example:
italian_clean_dataset = load_dataset("matteogabburo/mWikiQA", "it_clean")

Format:

Each example has the following format:

{
  'eid': 1214,
  'qid': 141,
  'cid': 0,
  'label': 1,
  'question': 'Was bedeutet Karma im Buddhismus?',
  'candidate': 'Karma (Sanskrit, auch karman, Pali: Kamma) bedeutet "Handlung" oder "Tun"; was auch immer man tut, sagt oder denkt, ist ein Karma.'
}

Where:

  • eid: is the unique id of the example (question, candidate)
  • qid: is the unique id of the question
  • cid: is the unique id of the answer candidate
  • label: identifies whether the answer candidate candidate is correct for the question (1 if correct, 0 otherwise)
  • question: the question
  • candidate: the answer candidate

Citation

If you find this dataset useful, please cite the following paper:

BibTeX:

@misc{gabburo2024datasetsmultilingualanswersentence,
      title={Datasets for Multilingual Answer Sentence Selection}, 
      author={Matteo Gabburo and Stefano Campese and Federico Agostini and Alessandro Moschitti},
      year={2024},
      eprint={2406.10172},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2406.10172}, 
}