SEACrowd/id_multilabel_hs

1from pathlib import Path2from typing import Dict, List, Tuple3 4import datasets5import pandas as pd6 7from seacrowd.utils import schemas8from seacrowd.utils.configs import SEACrowdConfig9from seacrowd.utils.constants import Tasks10 11_CITATION = """\12@inproceedings{ibrohim-budi-2019-multi,13    title = "Multi-label Hate Speech and Abusive Language Detection in {I}ndonesian {T}witter",14    author = "Ibrohim, Muhammad Okky  and15      Budi, Indra",16    booktitle = "Proceedings of the Third Workshop on Abusive Language Online",17    month = aug,18    year = "2019",19    address = "Florence, Italy",20    publisher = "Association for Computational Linguistics",21    url = "https://aclanthology.org/W19-3506",22    doi = "10.18653/v1/W19-3506",23    pages = "46--57",24}25"""26 27_LOCAL = False28_LANGUAGES = ["ind"]  # We follow ISO639-3 language code (https://iso639-3.sil.org/code_tables/639/data)29_DATASETNAME = "id_multilabel_hs"30 31_DESCRIPTION = """\32The ID_MULTILABEL_HS dataset is collection of 13,169 tweets in Indonesian language,33designed for hate speech detection NLP task. This dataset is combination from previous research and newly crawled data from Twitter.34This is a multilabel dataset with label details as follows:35-HS : hate speech label;36-Abusive : abusive language label;37-HS_Individual : hate speech targeted to an individual;38-HS_Group : hate speech targeted to a group;39-HS_Religion : hate speech related to religion/creed;40-HS_Race : hate speech related to race/ethnicity;41-HS_Physical : hate speech related to physical/disability;42-HS_Gender : hate speech related to gender/sexual orientation;43-HS_Gender : hate related to other invective/slander;44-HS_Weak : weak hate speech;45-HS_Moderate : moderate hate speech;46-HS_Strong : strong hate speech.47"""48 49_HOMEPAGE = "https://aclanthology.org/W19-3506/"50_LICENSE = "Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International"51_URLS = {52    _DATASETNAME: "https://raw.githubusercontent.com/okkyibrohim/id-multi-label-hate-speech-and-abusive-language-detection/master/re_dataset.csv",53}54_SUPPORTED_TASKS = [Tasks.ASPECT_BASED_SENTIMENT_ANALYSIS]55_SOURCE_VERSION = "1.0.0"56_SEACROWD_VERSION = "2024.06.20"57 58 59class IdAbusive(datasets.GeneratorBasedBuilder):60    """The ID_MULTILABEL_HS dataset is multi-label hate speech and abusive language detection in Indonesian tweets"""61 62    SOURCE_VERSION = datasets.Version(_SOURCE_VERSION)63    SEACROWD_VERSION = datasets.Version(_SEACROWD_VERSION)64 65    BUILDER_CONFIGS = [66        SEACrowdConfig(67            name="id_multilabel_hs_source",68            version=SOURCE_VERSION,69            description="ID Multilabel HS source schema",70            schema="source",71            subset_id="id_multilabel_hs",72        ),73        SEACrowdConfig(74            name="id_multilabel_hs_seacrowd_text_multi",75            version=SEACROWD_VERSION,76            description="ID Multilabel HS Nusantara schema",77            schema="seacrowd_text_multi",78            subset_id="id_multilabel_hs",79        ),80    ]81 82    DEFAULT_CONFIG_NAME = "id_multilabel_hs_source"83 84    def _info(self) -> datasets.DatasetInfo:85        if self.config.schema == "source":86            features = datasets.Features({87                "tweet": datasets.Value("string"), 88                "HS": datasets.Value("bool"),89                "Abusive": datasets.Value("bool"), 90                "HS_Individual": datasets.Value("bool"), 91                "HS_Group": datasets.Value("bool"), 92                "HS_Religion": datasets.Value("bool"), 93                "HS_Race": datasets.Value("bool"), 94                "HS_Physical": datasets.Value("bool"), 95                "HS_Gender": datasets.Value("bool"), 96                "HS_Other": datasets.Value("bool"), 97                "HS_Weak": datasets.Value("bool"), 98                "HS_Moderate": datasets.Value("bool"), 99                "HS_Strong": datasets.Value("bool"),100                })101        elif self.config.schema == "seacrowd_text_multi":102            features = schemas.text_multi_features([0, 1])103 104        return datasets.DatasetInfo(105            description=_DESCRIPTION,106            features=features,107            homepage=_HOMEPAGE,108            license=_LICENSE,109            citation=_CITATION,110        )111 112    def _split_generators(self, dl_manager: datasets.DownloadManager) -> List[datasets.SplitGenerator]:113        """Returns SplitGenerators."""114        # Dataset does not have predetermined split, putting all as TRAIN115        urls = _URLS[_DATASETNAME]116        base_dir = Path(dl_manager.download_and_extract(urls))117        data_files = {"train": base_dir}118 119        return [120            datasets.SplitGenerator(121                name=datasets.Split.TRAIN,122                gen_kwargs={123                    "filepath": data_files["train"],124                    "split": "train",125                },126            ),127        ]128 129    def _generate_examples(self, filepath: Path, split: str) -> Tuple[int, Dict]:130        """Yields examples as (key, example) tuples."""131        # Dataset does not have id, using row index as id132        label_cols = ["HS", "Abusive", "HS_Individual", "HS_Group", "HS_Religion", "HS_Race", "HS_Physical", "HS_Gender", "HS_Other", "HS_Weak", "HS_Moderate", "HS_Strong"]133        df = pd.read_csv(filepath, encoding="ISO-8859-1").reset_index()134        df.columns = ["id", "tweet"] + label_cols135 136        if self.config.schema == "source":137            for row in df.itertuples():138                ex = {139                    "tweet": row.tweet,140                }141                for label in label_cols:142                    ex[label] = getattr(row, label)143                yield row.id, ex144 145        elif self.config.schema == "seacrowd_text_multi":146            for row in df.itertuples():147                ex = {148                    "id": str(row.id),149                    "text": row.tweet,150                    "labels": [label for label in row[3:]],151                }152                yield row.id, ex153        else:154            raise ValueError(f"Invalid config: {self.config.name}")155