from pathlib import Path from typing import Dict, List, Tuple import datasets import pandas as pd from seacrowd.utils import schemas from seacrowd.utils.configs import SEACrowdConfig from seacrowd.utils.constants import Tasks _CITATION = """\ @article{karo2022sentiment, title={Sentiment Analysis in Karonese Tweet using Machine Learning}, author={Karo, Ichwanul Muslim Karo and Fudzee, Mohd Farhan Md and Kasim, Shahreen and Ramli, Azizul Azhar}, journal={Indonesian Journal of Electrical Engineering and Informatics (IJEEI)}, volume={10}, number={1}, pages={219--231}, year={2022} } """ _LANGUAGES = ["btx"] _LOCAL = False _DATASETNAME = "karonese_sentiment" _DESCRIPTION = """\ Karonese sentiment was crawled from Twitter between 1 January 2021 and 31 October 2021. The first crawling process used several keywords related to the Karonese, such as "deleng sinabung, Sinabung mountain", "mejuah-juah, greeting welcome", "Gundaling", and so on. However, due to the insufficient number of tweets obtained using such keywords, a second crawling process was done based on several hashtags, such as #kalakkaro, # #antonyginting, and #lyodra. """ _HOMEPAGE = "http://section.iaesonline.com/index.php/IJEEI/article/view/3565" _LICENSE = "Unknown" _URLS = { _DATASETNAME: "https://raw.githubusercontent.com/aliakbars/karonese/main/karonese_sentiment.csv", } _SUPPORTED_TASKS = [Tasks.SENTIMENT_ANALYSIS] _SOURCE_VERSION = "1.0.0" _SEACROWD_VERSION = "2024.06.20" class KaroneseSentimentDataset(datasets.GeneratorBasedBuilder): """Karonese sentiment was crawled from Twitter between 1 January 2021 and 31 October 2021. The dataset consists of 397 negative, 342 neutral, and 260 positive tweets. """ SOURCE_VERSION = datasets.Version(_SOURCE_VERSION) SEACROWD_VERSION = datasets.Version(_SEACROWD_VERSION) BUILDER_CONFIGS = [ SEACrowdConfig( name="karonese_sentiment_source", version=SOURCE_VERSION, description="Karonese Sentiment source schema", schema="source", subset_id="karonese_sentiment", ), SEACrowdConfig( name="karonese_sentiment_seacrowd_text", version=SEACROWD_VERSION, description="Karonese Sentiment Nusantara schema", schema="seacrowd_text", subset_id="karonese_sentiment", ), ] DEFAULT_CONFIG_NAME = "sentiment_nathasa_review_source" def _info(self) -> datasets.DatasetInfo: if self.config.schema == "source": features = datasets.Features( { "no": datasets.Value("string"), "tweet": datasets.Value("string"), "label": datasets.Value("string"), } ) elif self.config.schema == "seacrowd_text": features = schemas.text_features(["negative", "neutral", "positive"]) return datasets.DatasetInfo( description=_DESCRIPTION, features=features, homepage=_HOMEPAGE, license=_LICENSE, citation=_CITATION, ) def _split_generators(self, dl_manager: datasets.DownloadManager) -> List[datasets.SplitGenerator]: """Returns SplitGenerators.""" # Dataset does not have predetermined split, putting all as TRAIN data_dir = Path(dl_manager.download_and_extract(_URLS[_DATASETNAME])) return [ datasets.SplitGenerator( name=datasets.Split.TRAIN, gen_kwargs={ "filepath": data_dir, }, ), ] def _generate_examples(self, filepath: Path) -> Tuple[int, Dict]: """Yields examples as (key, example) tuples.""" df = pd.read_csv(filepath).drop("no", axis=1) df.columns = ["text", "label"] if self.config.schema == "source": for idx, row in df.iterrows(): example = { "no": str(idx+1), "tweet": row.text, "label": row.label, } yield idx, example elif self.config.schema == "seacrowd_text": for idx, row in df.iterrows(): example = { "id": str(idx+1), "text": row.text, "label": row.label, } yield idx, example else: raise ValueError(f"Invalid config: {self.config.name}")