File size: 3,964 Bytes
83d24b2 | 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 | from __future__ import annotations
import datasets
from mteb.abstasks.TaskMetadata import TaskMetadata
from ....abstasks import AbsTaskBitextMining, CrosslingualTask
_LANGUAGES = {
"ar-en": ["ara-Arab", "eng-Latn"],
"de-en": ["deu-Latn", "eng-Latn"],
"en-ar": ["eng-Latn", "ara-Arab"],
"en-de": ["eng-Latn", "deu-Latn"],
"en-fr": ["eng-Latn", "fra-Latn"],
"en-it": ["eng-Latn", "ita-Latn"],
"en-ja": ["eng-Latn", "jpn-Jpan"],
"en-ko": ["eng-Latn", "kor-Hang"],
"en-nl": ["eng-Latn", "nld-Latn"],
"en-ro": ["eng-Latn", "ron-Latn"],
"en-zh": ["eng-Latn", "cmn-Hans"],
"fr-en": ["fra-Latn", "eng-Latn"],
"it-en": ["ita-Latn", "eng-Latn"],
"it-nl": ["ita-Latn", "nld-Latn"],
"it-ro": ["ita-Latn", "ron-Latn"],
"ja-en": ["jpn-Jpan", "eng-Latn"],
"ko-en": ["kor-Hang", "eng-Latn"],
"nl-en": ["nld-Latn", "eng-Latn"],
"nl-it": ["nld-Latn", "ita-Latn"],
"nl-ro": ["nld-Latn", "ron-Latn"],
"ro-en": ["ron-Latn", "eng-Latn"],
"ro-it": ["ron-Latn", "ita-Latn"],
"ro-nl": ["ron-Latn", "nld-Latn"],
"zh-en": ["cmn-Hans", "eng-Latn"],
}
_SPLITS = ["validation"]
class IWSLT2017BitextMining(AbsTaskBitextMining, CrosslingualTask):
metadata = TaskMetadata(
name="IWSLT2017BitextMining",
dataset={
"path": "IWSLT/iwslt2017",
"revision": "c18a4f81a47ae6fa079fe9d32db288ddde38451d",
},
description="The IWSLT 2017 Multilingual Task addresses text translation, including zero-shot translation, with a single MT system across all directions including English, German, Dutch, Italian and Romanian.",
reference="https://aclanthology.org/2017.iwslt-1.1/",
type="BitextMining",
category="s2s",
eval_splits=_SPLITS,
eval_langs=_LANGUAGES,
main_score="f1",
date=("2007-01-01", "2017-12-14"), # rough estimate
form=["written"],
domains=["Non-fiction", "Fiction"],
task_subtypes=[],
license="CC-BY-NC-ND-4.0",
socioeconomic_status="medium",
annotations_creators="expert-annotated",
dialect=[],
text_creation="found",
bibtex_citation="""
@inproceedings{cettolo-etal-2017-overview,
title = "Overview of the {IWSLT} 2017 Evaluation Campaign",
author = {Cettolo, Mauro and
Federico, Marcello and
Bentivogli, Luisa and
Niehues, Jan and
St{\"u}ker, Sebastian and
Sudoh, Katsuhito and
Yoshino, Koichiro and
Federmann, Christian},
editor = "Sakti, Sakriani and
Utiyama, Masao",
booktitle = "Proceedings of the 14th International Conference on Spoken Language Translation",
month = dec # " 14-15",
year = "2017",
address = "Tokyo, Japan",
publisher = "International Workshop on Spoken Language Translation",
url = "https://aclanthology.org/2017.iwslt-1.1",
pages = "2--14",
}
""",
n_samples={"validation": 21928},
avg_character_length={"validation": 95.4},
)
def load_data(self, **kwargs):
"""Load dataset from HuggingFace hub and convert it to the standard format."""
if self.data_loaded:
return
self.dataset = {}
for lang in self.hf_subsets:
self.dataset[lang] = datasets.load_dataset(
split=_SPLITS,
name=f"iwslt2017-{lang}",
**self.metadata_dict["dataset"],
)
self.dataset_transform()
self.data_loaded = True
def dataset_transform(self):
def create_columns(row, lang):
l1, l2 = lang.split("-")
row["sentence1"] = row["translation"][l1]
row["sentence2"] = row["translation"][l2]
return row
# Convert to standard format
for lang in self.hf_subsets:
self.dataset[lang] = self.dataset[lang].map(
lambda x: create_columns(x, lang=lang)
)
|