|
from pathlib import Path |
|
from typing import Dict, List, Tuple |
|
|
|
import datasets |
|
|
|
from seacrowd.utils.configs import SEACrowdConfig |
|
from seacrowd.utils.constants import Tasks |
|
from seacrowd.utils import schemas |
|
|
|
import pandas as pd |
|
|
|
|
|
_CITATION = """\ |
|
@INPROCEEDINGS{8629151, |
|
author={Aliyah Salsabila, Nikmatun and Ardhito Winatmoko, Yosef and Akbar Septiandri, Ali and Jamal, Ade}, |
|
booktitle={2018 International Conference on Asian Language Processing (IALP)}, |
|
title={Colloquial Indonesian Lexicon}, |
|
year={2018}, |
|
volume={}, |
|
number={}, |
|
pages={226-229}, |
|
doi={10.1109/IALP.2018.8629151}} |
|
""" |
|
|
|
_LANGUAGES = ["ind"] |
|
_LOCAL = False |
|
|
|
_DATASETNAME = "kamus_alay" |
|
|
|
_DESCRIPTION = """\ |
|
Kamus Alay provide a lexicon for text normalization of Indonesian colloquial words. |
|
It contains 3,592 unique colloquial words-also known as “bahasa alay” -and manually annotated them |
|
with the normalized form. We built this lexicon from Instagram comments provided by Septiandri & Wibisono (2017) |
|
""" |
|
|
|
_HOMEPAGE = "https://ieeexplore.ieee.org/abstract/document/8629151" |
|
|
|
_LICENSE = "Unknown" |
|
|
|
_URLS = { |
|
_DATASETNAME: "https://raw.githubusercontent.com/nasalsabila/kamus-alay/master/colloquial-indonesian-lexicon.csv", |
|
} |
|
|
|
_SUPPORTED_TASKS = [Tasks.MORPHOLOGICAL_INFLECTION] |
|
|
|
|
|
_SOURCE_VERSION = "1.0.0" |
|
_SEACROWD_VERSION = "2024.06.20" |
|
|
|
|
|
class KamusAlay(datasets.GeneratorBasedBuilder): |
|
"""Kamus Alay is a dataset of lexicon for text normalization of Indonesian colloquial word""" |
|
|
|
SOURCE_VERSION = datasets.Version(_SOURCE_VERSION) |
|
SEACROWD_VERSION = datasets.Version(_SEACROWD_VERSION) |
|
|
|
label_classes = [ |
|
"abreviasi", |
|
"afiksasi", |
|
"akronim", |
|
"anaptiksis", |
|
"coinage", |
|
"elongasi", |
|
"homofon", |
|
"metatesis", |
|
"modifikasi vokal", |
|
"monoftongisasi", |
|
"naturalisasi", |
|
"pungtuasi", |
|
"reduplikasi", |
|
"salah ketik", |
|
"subtitusi", |
|
"word-value letter", |
|
"zeroisasi", |
|
] |
|
|
|
BUILDER_CONFIGS = [ |
|
SEACrowdConfig( |
|
name="kamus_alay_source", |
|
version=SOURCE_VERSION, |
|
description="Kamus Alay source schema", |
|
schema="source", |
|
subset_id="kamus_alay", |
|
), |
|
SEACrowdConfig( |
|
name="kamus_alay_seacrowd_pairs_multi", |
|
version=SEACROWD_VERSION, |
|
description="Kamus Alay Nusantara schema", |
|
schema="seacrowd_pairs_multi", |
|
subset_id="kamus_alay", |
|
), |
|
] |
|
|
|
DEFAULT_CONFIG_NAME = "kamus_alay_source" |
|
|
|
def _info(self) -> datasets.DatasetInfo: |
|
if self.config.schema == "source": |
|
features = datasets.Features( |
|
{ |
|
"slang": datasets.Value("string"), |
|
"formal": datasets.Value("string"), |
|
"in_dictionary": datasets.Value("bool"), |
|
"context": datasets.Value("string"), |
|
"categories": datasets.Sequence(datasets.Value("string")), |
|
} |
|
) |
|
elif self.config.schema == "seacrowd_pairs_multi": |
|
features = schemas.pairs_multi_features(self.label_classes) |
|
|
|
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.""" |
|
urls = _URLS[_DATASETNAME] |
|
|
|
data_dir = Path(dl_manager.download(urls)) |
|
return [ |
|
datasets.SplitGenerator( |
|
name=datasets.Split.TRAIN, |
|
gen_kwargs={ |
|
"filepath": data_dir, |
|
"split": "train", |
|
}, |
|
), |
|
] |
|
|
|
def _generate_examples(self, filepath: Path, split: str) -> Tuple[int, Dict]: |
|
"""Yields examples as (key, example) tuples.""" |
|
|
|
df = pd.read_csv(filepath, encoding="ISO-8859-1").reset_index() |
|
df.columns = ["id", "slang", "formal", "is_in_dictionary", "example", "category1", "category2", "category3"] |
|
|
|
if self.config.schema == "source": |
|
for row in df.itertuples(): |
|
ex = { |
|
"slang": row.slang, |
|
"formal": row.formal, |
|
"in_dictionary": row.is_in_dictionary, |
|
"context": row.example, |
|
"categories": [c for c in (row.category1, row.category2, row.category3) if c != "0"], |
|
} |
|
yield row.id, ex |
|
|
|
elif self.config.schema == "seacrowd_pairs_multi": |
|
for row in df.itertuples(): |
|
ex = { |
|
"id": str(row.id), |
|
"text_1": row.formal, |
|
"text_2": row.slang, |
|
"label": [c for c in (row.category1, row.category2, row.category3) if c != "0"], |
|
} |
|
yield row.id, ex |
|
else: |
|
raise ValueError(f"Invalid config: {self.config.name}") |
|
|