mind.py

import csv
import datasets

_TRAIN_DOWNLOAD_URL = "https://huggingface.co/datasets/linxinyuan/mind/resolve/main/train.csv"
_TEST_DOWNLOAD_URL = "https://huggingface.co/datasets/linxinyuan/mind/resolve/main/test.csv"

class mind(datasets.GeneratorBasedBuilder):
    def _info(self):
        return datasets.DatasetInfo(
            description="MIND",
            features=datasets.Features(
                {
                    "text": datasets.Value("string"),
                    "label": datasets.features.ClassLabel(names=['foodanddrink', 'games', 'lifestyle', 'weather', 'kids', 'entertainment', 'middleeast', 'news', 'autos', 'video', 'travel', 'music', 'northamerica', 'movies', 'finance', 'health', 'sports', 'tv']),
                }
            ),
        )

    def _split_generators(self, dl_manager):
        train_path = dl_manager.download_and_extract(_TRAIN_DOWNLOAD_URL)
        test_path = dl_manager.download_and_extract(_TEST_DOWNLOAD_URL)
        return [
            datasets.SplitGenerator(name=datasets.Split.TRAIN, gen_kwargs={"filepath": train_path}),
            datasets.SplitGenerator(name=datasets.Split.TEST, gen_kwargs={"filepath": test_path}),
        ]
        
    def _generate_examples(self, filepath):
        with open(filepath, encoding="utf-8") as csv_file:
            csv_reader = csv.reader(csv_file, delimiter="\t")
            for id_, row in enumerate(csv_reader):
                yield id_, {"text": row[3], "label": (int)(row[1])}