Datasets:
Tasks:
Translation
Multilinguality:
translation
Size Categories:
1M<n<10M
Source Datasets:
original
ArXiv:
Tags:
License:
File size: 4,013 Bytes
5af0d18 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 |
import json
import os
import datasets
_CITATION = """\
@article{lowphansirikul2020scb,
title={scb-mt-en-th-2020: A Large English-Thai Parallel Corpus},
author={Lowphansirikul, Lalita and Polpanumas, Charin and Rutherford, Attapol T and Nutanong, Sarana},
journal={arXiv preprint arXiv:2007.03541},
year={2020}
}
"""
_DESCRIPTION = """\
scb-mt-en-th-2020: A Large English-Thai Parallel Corpus
The primary objective of our work is to build a large-scale English-Thai dataset for machine translation.
We construct an English-Thai machine translation dataset with over 1 million segment pairs, curated from various sources,
namely news, Wikipedia articles, SMS messages, task-based dialogs, web-crawled data and government documents.
Methodology for gathering data, building parallel texts and removing noisy sentence pairs are presented in a reproducible manner.
We train machine translation models based on this dataset. Our models' performance are comparable to that of
Google Translation API (as of May 2020) for Thai-English and outperform Google when the Open Parallel Corpus (OPUS) is
included in the training data for both Thai-English and English-Thai translation.
The dataset, pre-trained models, and source code to reproduce our work are available for public use.
"""
class ScbMtEnth2020Config(datasets.BuilderConfig):
"""BuilderConfig for ScbMtEnth2020."""
def __init__(self, language_pair=(None, None), **kwargs):
"""BuilderConfig for ScbMtEnth2020.
Args:
**kwargs: keyword arguments forwarded to super.
"""
super(ScbMtEnth2020Config, self).__init__(
name=f"{language_pair[0]}{language_pair[1]}",
description="Translate {language_pair[0]} to {language_pair[1]}",
version=datasets.Version("1.0.0"),
**kwargs,
)
self.language_pair = language_pair
class ScbMtEnth2020(datasets.GeneratorBasedBuilder):
"""scb-mt-en-th-2020: A Large English-Thai Parallel Corpus"""
_DOWNLOAD_URL = "https://archive.org/download/scb_mt_enth_2020/data.zip"
_TRAIN_FILE = "train.jsonl"
_VAL_FILE = "valid.jsonl"
_TEST_FILE = "test.jsonl"
BUILDER_CONFIG_CLASS = ScbMtEnth2020Config
BUILDER_CONFIGS = [
ScbMtEnth2020Config(
language_pair=("en", "th"),
),
ScbMtEnth2020Config(
language_pair=("th", "en"),
),
]
def _info(self):
return datasets.DatasetInfo(
description=_DESCRIPTION,
features=datasets.Features(
{
"translation": datasets.features.Translation(languages=self.config.language_pair),
"subdataset": datasets.Value("string"),
}
),
supervised_keys=None,
homepage="https://airesearch.in.th/",
citation=_CITATION,
)
def _split_generators(self, dl_manager):
arch_path = dl_manager.download_and_extract(self._DOWNLOAD_URL)
data_dir = os.path.join(arch_path, "data")
return [
datasets.SplitGenerator(
name=datasets.Split.TRAIN, gen_kwargs={"filepath": os.path.join(data_dir, self._TRAIN_FILE)}
),
datasets.SplitGenerator(
name=datasets.Split.VALIDATION, gen_kwargs={"filepath": os.path.join(data_dir, self._VAL_FILE)}
),
datasets.SplitGenerator(
name=datasets.Split.TEST, gen_kwargs={"filepath": os.path.join(data_dir, self._TEST_FILE)}
),
]
def _generate_examples(self, filepath):
"""Generate examples."""
source, target = self.config.language_pair
with open(filepath, encoding="utf-8") as f:
for id_, row in enumerate(f):
data = json.loads(row)
yield id_, {
"translation": {source: data[source], target: data[target]},
"subdataset": data["subdataset"],
}
|