kpwr-long.py

# coding=utf-8
"""KPWR version 1.27 dataset."""

import csv
import datasets

_DESCRIPTION = "KPWR version 1.27 dataset. Prepared for Longformer."

_URLS = {
    "train": "https://huggingface.co/datasets/clarin-knext/kpwr-long/resolve/main/data/train.iob",
    "valid": "https://huggingface.co/datasets/clarin-knext/kpwr-long/resolve/main/data/valid.iob",
    "test": "https://huggingface.co/datasets/clarin-knext/kpwr-long/resolve/main/data/test.iob",
}

_HOMEPAGE = "https://clarin-pl.eu/dspace/handle/11321/270"

_N82_TAGS = [
    'nam_adj',
    'nam_adj_city',
    'nam_adj_country',
    'nam_adj_person',
    'nam_eve',
    'nam_eve_human',
    'nam_eve_human_cultural',
    'nam_eve_human_holiday',
    'nam_eve_human_sport',
    'nam_fac_bridge',
    'nam_fac_goe',
    'nam_fac_goe_stop',
    'nam_fac_park',
    'nam_fac_road',
    'nam_fac_square',
    'nam_fac_system',
    'nam_liv_animal',
    'nam_liv_character',
    'nam_liv_god',
    'nam_liv_habitant',
    'nam_liv_person',
    'nam_loc',
    'nam_loc_astronomical',
    'nam_loc_country_region',
    'nam_loc_gpe_admin1',
    'nam_loc_gpe_admin2',
    'nam_loc_gpe_admin3',
    'nam_loc_gpe_city',
    'nam_loc_gpe_conurbation',
    'nam_loc_gpe_country',
    'nam_loc_gpe_district',
    'nam_loc_gpe_subdivision',
    'nam_loc_historical_region',
    'nam_loc_hydronym',
    'nam_loc_hydronym_lake',
    'nam_loc_hydronym_ocean',
    'nam_loc_hydronym_river',
    'nam_loc_hydronym_sea',
    'nam_loc_land',
    'nam_loc_land_continent',
    'nam_loc_land_island',
    'nam_loc_land_mountain',
    'nam_loc_land_peak',
    'nam_loc_land_region',
    'nam_num_house',
    'nam_num_phone',
    'nam_org_company',
    'nam_org_group',
    'nam_org_group_band',
    'nam_org_group_team',
    'nam_org_institution',
    'nam_org_nation',
    'nam_org_organization',
    'nam_org_organization_sub',
    'nam_org_political_party',
    'nam_oth',
    'nam_oth_currency',
    'nam_oth_data_format',
    'nam_oth_license',
    'nam_oth_position',
    'nam_oth_tech',
    'nam_oth_www',
    'nam_pro',
    'nam_pro_award',
    'nam_pro_brand',
    'nam_pro_media',
    'nam_pro_media_periodic',
    'nam_pro_media_radio',
    'nam_pro_media_tv',
    'nam_pro_media_web',
    'nam_pro_model_car',
    'nam_pro_software',
    'nam_pro_software_game',
    'nam_pro_title',
    'nam_pro_title_album',
    'nam_pro_title_article',
    'nam_pro_title_book',
    'nam_pro_title_document',
    'nam_pro_title_song',
    'nam_pro_title_treaty',
    'nam_pro_title_tv',
    'nam_pro_vehicle'
]

_NER_IOB_TAGS = ['O']

for tag in _N82_TAGS:
    _NER_IOB_TAGS.extend([f'B-{tag}', f'I-{tag}'])


class KpwrDataset(datasets.GeneratorBasedBuilder):

    def _info(self) -> datasets.DatasetInfo:
        return datasets.DatasetInfo(
            description=_DESCRIPTION,
            features=datasets.Features(
                {
                    "tokens": datasets.Sequence(datasets.Value('string')),
                    "lemmas": datasets.Sequence(datasets.Value('string')),
                    "mstags": datasets.Sequence(datasets.Value('string')),
                    "ner": datasets.Sequence(datasets.features.ClassLabel(names=_NER_IOB_TAGS))
                }
            ),
            homepage=_HOMEPAGE
        )

    def _split_generators(self, dl_manager: datasets.DownloadManager):
        downloaded_files = dl_manager.download_and_extract(_URLS)
        return [
            datasets.SplitGenerator(name=datasets.Split.TRAIN, gen_kwargs={'filepath': downloaded_files['train']}),
            datasets.SplitGenerator(name=datasets.Split.VALIDATION, gen_kwargs={'filepath': downloaded_files['valid']}),
            datasets.SplitGenerator(name=datasets.Split.TEST, gen_kwargs={'filepath': downloaded_files['test']})
        ]

    def _generate_examples(self, filepath: str):
        with open(filepath, 'r', encoding='utf-8') as fin:
            reader = csv.reader(fin, delimiter='\t', quoting=csv.QUOTE_NONE)

            tokens = []
            lemmas = []
            mstags = []
            ner = []
            gid = 0

            for line in reader:
                if not line:
                    yield gid, {
                        "tokens": tokens,
                        "lemmas": lemmas,
                        "mstags": mstags,
                        "ner": ner
                    }
                    gid += 1
                    tokens = []
                    lemmas = []
                    mstags = []
                    ner = []
                
                elif len(line) == 1:  # ignore --DOCSTART lines
                    continue

                else:
                    tokens.append(line[0])
                    lemmas.append(line[1])
                    mstags.append(line[2])
                    ner.append(line[3])