JaQuAD.py

'''Dataset loading script for JaQuAD.
We refer to https://huggingface.co/datasets/squad_v2/blob/main/squad_v2.py
'''
import json
import os

import datasets

_CITATION = '''
@article{SkelterLabsInc:JaQuAD,
 title = {{JaQuAD}: Japanese Question Answering Dataset for Machine
          Reading Comprehension},
author = {Byunghoon So  and
          Kyuhong Byun  and
          Kyungwon Kang  and
          Seongjin Cho},
  year = {2022},
}
'''
_DESCRIPTION = '''Japanese Question Answering Dataset (JaQuAD), released in
2022, is a human-annotated dataset created for Japanese Machine Reading
Comprehension. JaQuAD is developed to provide a SQuAD-like QA dataset in
Japanese. JaQuAD contains 39,696 question-answer pairs. Questions and answers
are manually curated by human annotators. Contexts are collected from Japanese
Wikipedia articles.
'''
_LICENSE = 'CC BY-SA 3.0'
_HOMEPAGE = 'https://skelterlabs.com/en/'
_URL = 'https://huggingface.co/datasets/SkelterLabsInc/JaQuAD/raw/main/data/'


class JaQuAD(datasets.GeneratorBasedBuilder):
    VERSION = datasets.Version('0.1.0')

    def _info(self):
        features = datasets.Features({
            'id': datasets.Value('string'),
            'title': datasets.Value('string'),
            'context': datasets.Value('string'),
            'question': datasets.Value('string'),
            'question_type': datasets.Value('string'),
            'answers':
                datasets.features.Sequence({
                    'text': datasets.Value('string'),
                    'answer_start': datasets.Value('int32'),
                    'answer_type': datasets.Value('string'),
                }),
        })
        return datasets.DatasetInfo(
            description=_DESCRIPTION,
            features=features,
            homepage=_HOMEPAGE,
            license=_LICENSE,
            citation=_CITATION,
        )

    def _split_generators(self, dl_manager):
        urls_to_download = {
            'train': [
                os.path.join(_URL, f'train/jaquad_train_{i:04d}.json')
                for i in range(30)
            ],
            'dev': [
                os.path.join(_URL, f'dev/jaquad_dev_{i:04d}.json')
                for i in range(4)
            ],
        }
        downloaded_files = dl_manager.download_and_extract(urls_to_download)
        return [
            datasets.SplitGenerator(
                name=datasets.Split.TRAIN,
                gen_kwargs={'filepaths': downloaded_files['train']},
            ),
            datasets.SplitGenerator(
                name=datasets.Split.VALIDATION,
                gen_kwargs={'filepaths': downloaded_files['dev']},
            ),
        ]

    def _generate_examples(self, filepaths):
        for filename in filepaths:
            with open(filename, encoding='utf-8') as ifile:
                jaquad = json.load(ifile)
                for article in jaquad['data']:
                    title = article.get('title', '').strip()
                    for paragraph in article['paragraphs']:
                        context = paragraph['context'].strip()
                        for qa in paragraph['qas']:
                            qa_id = qa['id']
                            question = qa['question'].strip()
                            question_type = qa['question_type']

                            answer_starts = [
                                answer['answer_start']
                                for answer in qa['answers']
                            ]
                            answer_texts = [
                                answer['text'].strip()
                                for answer in qa['answers']
                            ]
                            answer_types = [
                                answer['answer_type']
                                for answer in qa['answers']
                            ]

                            yield qa_id, {
                                'title': title,
                                'context': context,
                                'question': question,
                                'question_type': question_type,
                                'id': qa_id,
                                'answers': {
                                    'text': answer_texts,
                                    'answer_start': answer_starts,
                                    'answer_type': answer_types,
                                },
                            }