File size: 5,288 Bytes
4f0c1e0
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
73a9965
 
 
4f0c1e0
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1b7f1de
4f0c1e0
 
 
 
1b7f1de
 
 
 
 
4f0c1e0
 
 
 
1b7f1de
4f0c1e0
1b7f1de
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
# coding=utf-8
# Copyright 2020 The TensorFlow Datasets Authors and the HuggingFace Datasets Authors.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

# Lint as: python3
"""MOCHA: A Dataset for Training and Evaluating Generative Reading Comprehension Metrics"""


import json

import datasets


logger = datasets.logging.get_logger(__name__)


_CITATION = """\
@inproceedings{Chen2020MOCHAAD,
    author={Anthony Chen and Gabriel Stanovsky and Sameer Singh and Matt Gardner},
    title={MOCHA: A Dataset for Training and Evaluating Generative Reading Comprehension Metrics},
    booktitle={EMNLP},
    year={2020}
}
"""

_DESCRIPTION = """\
Posing reading comprehension as a generation problem provides a great deal of flexibility, allowing for \
open-ended questions with few restrictions on possible answers. However, progress is impeded by existing \
generation metrics, which rely on token overlap and are agnostic to the nuances of reading comprehension. \
To address this, we introduce a benchmark for training and evaluating generative reading comprehension metrics: \
MOdeling Correctness with Human Annotations. MOCHA contains 40K human judgement scores on model outputs from \
6 diverse question answering datasets and an additional set of minimal pairs for evaluation. Using MOCHA, \
we train an evaluation metric: LERC, a Learned Evaluation metric for Reading Comprehension, to mimic human \
judgement scores.
"""

_HOMEPAGE = "https://allennlp.org/mocha"

_LICENSE = "https://creativecommons.org/licenses/by-sa/4.0/legalcode"

_URL = "https://github.com/anthonywchen/MOCHA/raw/main/data/mocha.tar.gz"

_MINIMAL_PAIRS_SPLIT = "minimal_pairs"

SPLIT_FILENAMES = {
    datasets.Split.TRAIN: "train.json",
    datasets.Split.VALIDATION: "dev.json",
    datasets.Split.TEST: "test_no_labels.json",
    _MINIMAL_PAIRS_SPLIT: "minimal_pairs.json",
}


class Mocha(datasets.GeneratorBasedBuilder):
    """MOCHA: A Dataset for Training and Evaluating Generative Reading Comprehension Metrics"""

    def _info(self):
        return datasets.DatasetInfo(
            description=_DESCRIPTION,
            features=datasets.Features(
                {
                    "constituent_dataset": datasets.Value("string"),
                    "id": datasets.Value("string"),
                    "context": datasets.Value("string"),
                    "question": datasets.Value("string"),
                    "reference": datasets.Value("string"),
                    "candidate": datasets.Value("string"),
                    "score": datasets.Value("float"),
                    "metadata": {
                        "scores": datasets.features.Sequence(datasets.Value("int32")),
                        "source": datasets.Value("string"),
                    },
                    # features for minimal pairs
                    "candidate2": datasets.Value("string"),
                    "score2": datasets.Value("float"),
                }
            ),
            supervised_keys=None,
            homepage=_HOMEPAGE,
            license=_LICENSE,
            citation=_CITATION,
        )

    def _split_generators(self, dl_manager):
        archive = dl_manager.download(_URL)

        return [
            datasets.SplitGenerator(
                name=split,
                gen_kwargs={
                    "filepath": "mocha/" + SPLIT_FILENAMES[split],
                    "split": split,
                    "files": dl_manager.iter_archive(archive),
                },
            )
            for split in SPLIT_FILENAMES
        ]

    def _generate_examples(self, filepath, split, files):
        """This function returns the examples in the raw (text) form."""
        for path, f in files:
            if path == filepath:
                mocha = json.load(f)
                for constituent_dataset, samples in mocha.items():
                    for id_, sample in samples.items():
                        sample["id"] = id_
                        sample["constituent_dataset"] = constituent_dataset

                        # Add default values
                        if split == _MINIMAL_PAIRS_SPLIT:
                            sample["candidate"] = sample["candidate1"]
                            sample["score"] = sample["score1"]
                            del sample["candidate1"], sample["score1"]
                            sample["metadata"] = {"scores": [], "source": ""}
                        else:
                            if "score" not in sample:
                                sample["score"] = -1.0
                                sample["metadata"]["scores"] = []
                            sample["candidate2"] = ""
                            sample["score2"] = -1.0

                        yield id_, sample
                break