clefip2011.py

import os
import pandas as pd
import datasets


class CLEFIP2011Config(datasets.BuilderConfig):
    """Custom Config for CLEFIP2011"""

    def __init__(self, dataset_type=None, **kwargs):
        super(CLEFIP2011Config, self).__init__(**kwargs)
        self.dataset_type = dataset_type


class CLEFIP2011(datasets.GeneratorBasedBuilder):
    """Custom Dataset Loader"""

    BUILDER_CONFIGS = [
        CLEFIP2011Config(
            name="bibliographic",
            version=datasets.Version("1.0.0"),
            description="CLEF-IP 2011 Bibliographic Data",
            dataset_type="bibliographic",
        ),
    ]

    def _info(self):

        if self.config.dataset_type == "bibliographic":
            features = datasets.Features(
                {
                    "ucid": datasets.Value("string"),
                    "country": datasets.Value("string"),
                    "doc_number": datasets.Value("string"),
                    "kind": datasets.Value("string"),
                    "lang": datasets.Value("string"),
                    "corrected_lang": datasets.Value("string"),
                    "date": datasets.Value("string"),
                    "family_id": datasets.Value("string"),
                    "date_produced": datasets.Value("string"),
                    "status": datasets.Value("string"),
                    "ecla_list": datasets.Value("string"),
                    "applicant_name_list": datasets.Value("string"),
                    "inventor_name_list": datasets.Value("string"),
                    "title_de_text": datasets.Value("string"),
                    "title_fr_text": datasets.Value("string"),
                    "title_en_text": datasets.Value("string"),
                    "abstract_de_exist": datasets.Value("bool"),
                    "abstract_fr_exist": datasets.Value("bool"),
                    "abstract_en_exist": datasets.Value("bool"),
                    "description_de_exist": datasets.Value("bool"),
                    "description_fr_exist": datasets.Value("bool"),
                    "description_en_exist": datasets.Value("bool"),
                    "claims_de_exist": datasets.Value("bool"),
                    "claims_fr_exist": datasets.Value("bool"),
                    "claims_en_exist": datasets.Value("bool"),
                }
            )
        return datasets.DatasetInfo(
            description="CLEF-IP 2011 Bibliographic dataset.",
            features=features,
            supervised_keys=None,
        )

    def _split_generators(self, dl_manager):

        archive_path = dl_manager.download_and_extract(
            "https://huggingface.co./datasets/amylonidis/PatClass2011/resolve/main/clefip2011_bibliographic_clean.tar.gz"
        )

        bibliographic_file = os.path.join(archive_path, "clefip2011_bibliographic.csv")

        return [
            datasets.SplitGenerator(
                name=datasets.Split.TRAIN,
                gen_kwargs={
                    "filepaths": [bibliographic_file],
                    "split": "train",
                },
            ),
        ]

    def _generate_examples(self, filepaths, split):

        for filepath in filepaths:
            df = pd.read_csv(filepath, header=None)

            column_names = [
                "ucid",
                "country",
                "doc_number",
                "kind",
                "lang",
                "corrected_lang",
                "date",
                "family_id",
                "date_produced",
                "status",
                "ecla_list",
                "applicant_name_list",
                "inventor_name_list",
                "title_de_text",
                "title_fr_text",
                "title_en_text",
                "abstract_de_exist",
                "abstract_fr_exist",
                "abstract_en_exist",
                "description_de_exist",
                "description_fr_exist",
                "description_en_exist",
                "claims_de_exist",
                "claims_fr_exist",
                "claims_en_exist",
            ]
            df.columns = column_names

            df["date"] = pd.to_datetime(df["date"], format="%Y%m%d").astype(str)

            df["date_produced"] = pd.to_datetime(
                df["date_produced"], format="%Y%m%d"
            ).astype(str)

            boolean_columns = [
                "abstract_de_exist",
                "abstract_fr_exist",
                "abstract_en_exist",
                "description_de_exist",
                "description_fr_exist",
                "description_en_exist",
                "claims_de_exist",
                "claims_fr_exist",
                "claims_en_exist",
            ]
            for col in boolean_columns:
                df[col] = df[col].astype(bool)

            for idx, row in df.iterrows():
                yield idx, row.to_dict()