Datasets:

Sub-tasks:
extractive-qa
Multilinguality:
multilingual
Size Categories:
10K<n<100K
Language Creators:
crowdsourced
Annotations Creators:
crowdsourced
Source Datasets:
original
Tags:
License:
File size: 8,438 Bytes
29fac8c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
7ffe9b1
29fac8c
7ffe9b1
29fac8c
7ffe9b1
29fac8c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
58040f3
29fac8c
 
 
 
 
 
58040f3
 
29fac8c
 
 
 
 
 
58040f3
 
29fac8c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
58040f3
29fac8c
 
 
 
 
 
 
 
58040f3
29fac8c
 
 
 
 
 
58040f3
29fac8c
 
 
 
 
 
58040f3
 
29fac8c
 
 
 
58040f3
29fac8c
58040f3
 
 
 
 
 
 
 
29fac8c
 
 
 
 
 
 
 
 
 
 
7ffe9b1
 
 
29fac8c
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
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
"""TODO(mlqa): Add a description here."""


import json
import os

import datasets


# TODO(mlqa): BibTeX citation
_CITATION = """\
@article{lewis2019mlqa,
  title={MLQA: Evaluating Cross-lingual Extractive Question Answering},
  author={Lewis, Patrick and Oguz, Barlas and Rinott, Ruty and Riedel, Sebastian and Schwenk, Holger},
  journal={arXiv preprint arXiv:1910.07475},
  year={2019}
}
"""

# TODO(mlqa):
_DESCRIPTION = """\
    MLQA (MultiLingual Question Answering) is a benchmark dataset for evaluating cross-lingual question answering performance.
    MLQA consists of over 5K extractive QA instances (12K in English) in SQuAD format in seven languages - English, Arabic,
    German, Spanish, Hindi, Vietnamese and Simplified Chinese. MLQA is highly parallel, with QA instances parallel between
    4 different languages on average.
"""
_URL = "https://dl.fbaipublicfiles.com/MLQA/"
_DEV_TEST_URL = "MLQA_V1.zip"
_TRANSLATE_TEST_URL = "mlqa-translate-test.tar.gz"
_TRANSLATE_TRAIN_URL = "mlqa-translate-train.tar.gz"
_LANG = ["ar", "de", "vi", "zh", "en", "es", "hi"]
_TRANSLATE_LANG = ["ar", "de", "vi", "zh", "es", "hi"]


class MlqaConfig(datasets.BuilderConfig):
    def __init__(self, data_url, **kwargs):
        """BuilderConfig for MLQA

        Args:
          data_url: `string`, url to the dataset
          **kwargs: keyword arguments forwarded to super.
        """
        super(MlqaConfig, self).__init__(
            version=datasets.Version(
                "1.0.0",
            ),
            **kwargs,
        )
        self.data_url = data_url


class Mlqa(datasets.GeneratorBasedBuilder):
    """TODO(mlqa): Short description of my dataset."""

    # TODO(mlqa): Set up version.
    VERSION = datasets.Version("1.0.0")
    BUILDER_CONFIGS = (
        [
            MlqaConfig(
                name="mlqa-translate-train." + lang,
                data_url=_URL + _TRANSLATE_TRAIN_URL,
                description="Machine-translated data for Translate-train (SQuAD Train and Dev sets machine-translated into "
                "Arabic, German, Hindi, Vietnamese, Simplified Chinese and Spanish)",
            )
            for lang in _LANG
            if lang != "en"
        ]
        + [
            MlqaConfig(
                name="mlqa-translate-test." + lang,
                data_url=_URL + _TRANSLATE_TEST_URL,
                description="Machine-translated data for Translate-Test (MLQA-test set machine-translated into English) ",
            )
            for lang in _LANG
            if lang != "en"
        ]
        + [
            MlqaConfig(
                name="mlqa." + lang1 + "." + lang2,
                data_url=_URL + _DEV_TEST_URL,
                description="development and test splits",
            )
            for lang1 in _LANG
            for lang2 in _LANG
        ]
    )

    def _info(self):
        # TODO(mlqa): Specifies the datasets.DatasetInfo object
        return datasets.DatasetInfo(
            # This is the description that will appear on the datasets page.
            description=_DESCRIPTION,
            # datasets.features.FeatureConnectors
            features=datasets.Features(
                {
                    "context": datasets.Value("string"),
                    "question": datasets.Value("string"),
                    "answers": datasets.features.Sequence(
                        {"answer_start": datasets.Value("int32"), "text": datasets.Value("string")}
                    ),
                    "id": datasets.Value("string"),
                    # These are the features of your dataset like images, labels ...
                }
            ),
            # If there's a common (input, target) tuple from the features,
            # specify them here. They'll be used if as_supervised=True in
            # builder.as_dataset.
            supervised_keys=None,
            # Homepage of the dataset for documentation
            homepage="https://github.com/facebookresearch/MLQA",
            citation=_CITATION,
        )

    def _split_generators(self, dl_manager):
        """Returns SplitGenerators."""
        # TODO(mlqa): Downloads the data and defines the splits
        # dl_manager is a datasets.download.DownloadManager that can be used to
        # download and extract URLs
        if self.config.name.startswith("mlqa-translate-train"):
            archive = dl_manager.download(self.config.data_url)
            lang = self.config.name.split(".")[-1]
            return [
                datasets.SplitGenerator(
                    name=datasets.Split.TRAIN,
                    # These kwargs will be passed to _generate_examples
                    gen_kwargs={
                        "filepath": f"mlqa-translate-train/{lang}_squad-translate-train-train-v1.1.json",
                        "files": dl_manager.iter_archive(archive),
                    },
                ),
                datasets.SplitGenerator(
                    name=datasets.Split.VALIDATION,
                    # These kwargs will be passed to _generate_examples
                    gen_kwargs={
                        "filepath": f"mlqa-translate-train/{lang}_squad-translate-train-dev-v1.1.json",
                        "files": dl_manager.iter_archive(archive),
                    },
                ),
            ]

        else:
            if self.config.name.startswith("mlqa."):
                dl_file = dl_manager.download_and_extract(self.config.data_url)
                name = self.config.name.split(".")
                l1, l2 = name[1:]
                return [
                    datasets.SplitGenerator(
                        name=datasets.Split.TEST,
                        # These kwargs will be passed to _generate_examples
                        gen_kwargs={
                            "filepath": os.path.join(
                                os.path.join(dl_file, "MLQA_V1/test"),
                                f"test-context-{l1}-question-{l2}.json",
                            )
                        },
                    ),
                    datasets.SplitGenerator(
                        name=datasets.Split.VALIDATION,
                        # These kwargs will be passed to _generate_examples
                        gen_kwargs={
                            "filepath": os.path.join(
                                os.path.join(dl_file, "MLQA_V1/dev"), f"dev-context-{l1}-question-{l2}.json"
                            )
                        },
                    ),
                ]
            else:
                if self.config.name.startswith("mlqa-translate-test"):
                    archive = dl_manager.download(self.config.data_url)
                    lang = self.config.name.split(".")[-1]
                    return [
                        datasets.SplitGenerator(
                            name=datasets.Split.TEST,
                            # These kwargs will be passed to _generate_examples
                            gen_kwargs={
                                "filepath": f"mlqa-translate-test/translate-test-context-{lang}-question-{lang}.json",
                                "files": dl_manager.iter_archive(archive),
                            },
                        ),
                    ]

    def _generate_examples(self, filepath, files=None):
        """Yields examples."""
        if self.config.name.startswith("mlqa-translate"):
            for path, f in files:
                if path == filepath:
                    data = json.loads(f.read().decode("utf-8"))
                    break
        else:
            with open(filepath, encoding="utf-8") as f:
                data = json.load(f)
        for examples in data["data"]:
            for example in examples["paragraphs"]:
                context = example["context"]
                for qa in example["qas"]:
                    question = qa["question"]
                    id_ = qa["id"]
                    answers = qa["answers"]
                    answers_start = [answer["answer_start"] for answer in answers]
                    answers_text = [answer["text"] for answer in answers]
                    yield id_, {
                        "context": context,
                        "question": question,
                        "answers": {"answer_start": answers_start, "text": answers_text},
                        "id": id_,
                    }