quail.py

import os
import json
import datasets
from datasets import BuilderConfig, Features, Value, Sequence


_DESCRIPTION = """
# 한국어 지시학습 데이터셋
- quail 데이터셋을 한국어로 변역한 데이터셋
"""

_CITATION = """
@inproceedings{KITD,
  title={언어 번역 모델을 통한 한국어 지시 학습 데이터 세트 구축},
  author={임영서, 추현창, 김산, 장진예, 정민영, 신사임},
  booktitle={제 35회 한글 및 한국어 정보처리 학술대회},
  pages={591--595},
  year={2023}
}
@inproceedings{KITD,
  title={Korean Instruction Tuning Dataset},
  author={Yeongseo Lim, HyeonChang Chu, San Kim, Jin Yea Jang, Minyoung Jung, Saim Shin},
  booktitle={Proceedings of the 35th Annual Conference on Human and Cognitive Language Technology},
  pages={591--595},
  year={2023}
}
"""

def _list(data_list):
    result = list()
    for data in data_list:
        result.append(data)
    return result

# quail
_QUAIL_FEATURES = Features({
    "data_index_by_user": Value(dtype="int32"),
    "id": Value(dtype="string"),
    "context_id": Value(dtype="string"),
    "question_id": Value(dtype="string"),
    "domain": Value(dtype="string"),
    "metadata": {
        "author": Value(dtype="string"),
        "title": Value(dtype="string"),
        "url": Value(dtype="string"),
    },
    "context": Value(dtype="string"),
    "question": Value(dtype="string"),
    "question_type": Value(dtype="string"),
    "answers": Sequence(Value(dtype="string")),
    "correct_answer_id": Value(dtype="int32"),
})

def _parsing_quail(file_path):
    with open(file_path, mode="r") as f:
        dataset = json.load(f)
    for _i, data in enumerate(dataset):
        _data_index_by_user = data["data_index_by_user"]
        _id = data["id"]
        _context_id = data["context_id"]
        _question_id = data["question_id"]
        _domain = data["domain"]
        _metadata = {
            "author": data["metadata"]["author"],
            "title": data["metadata"]["title"],
            "url": data["metadata"]["url"]
        }
        _context = data["context"]
        _question = data["question"]
        _question_type = data["question_type"]
        _answers = _list(data["_answers"])
        _correct_answer_id = data["correct_answer_id"]
        
        yield _i, {
            "data_index_by_user": _data_index_by_user,
            "id": _id,
            "context_id": _context_id,
            "question_id": _question_id,
            "domain": _domain,
            "metadata": _metadata,
            "context": _context,
            "question": _question,
            "question_type": _question_type,
            "answers": _answers,
            "correct_answer_id": _correct_answer_id,
        }

class QuailConfig(BuilderConfig):
    def __init__(self, name, feature, reading_fn, parsing_fn, citation, **kwargs):
        super(QuailConfig, self).__init__(
            name = name,
            version=datasets.Version("1.0.0"),
            **kwargs)
        self.feature = feature
        self.reading_fn = reading_fn
        self.parsing_fn = parsing_fn
        self.citation = citation

class QUAIL(datasets.GeneratorBasedBuilder):
    BUILDER_CONFIGS = [
        QuailConfig(
            name = "base",
            data_dir = "./quail",
            feature = _QUAIL_FEATURES,
            reading_fn = _parsing_quail,
            parsing_fn = lambda x:x,
            citation = _CITATION,
        ),
    ]
    
    def _info(self) -> datasets.DatasetInfo:
        """Returns the dataset metadata."""
        return datasets.DatasetInfo(
            description=_DESCRIPTION,
            features=_QUAIL_FEATURES,
            citation=_CITATION,
        )
    
    def _split_generators(self, dl_manager: datasets.DownloadManager):
        """Returns SplitGenerators"""
        path_kv = {
            datasets.Split.TRAIN:[
                os.path.join(dl_manager.manual_dir, f"train.json")
            ],
            datasets.Split.VALIDATION:[
                os.path.join(dl_manager.manual_dir, f"validation.json")
            ],
            "challenge":[
                os.path.join(dl_manager.manual_dir, f"challenge.json")
            ],
        }
        return [
            datasets.SplitGenerator(name=k, gen_kwargs={"path_list": v})
            for k, v in path_kv.items()
        ]
    
    def _generate_examples(self, path_list):
        """Yields examples."""
        for path in path_list:
            try:
                for example in iter(self.config.reading_fn(path)):
                    yield self.config.parsing_fn(example)
            except Exception as e:
                print(e)