asnq.py

# coding=utf-8
# Copyright 2020 The HuggingFace Datasets Authors and the current dataset script contributor.
#
# 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.
"""Answer-Sentence Natural Questions (ASNQ)

ASNQ is a dataset for answer sentence selection derived from Google's
Natural Questions (NQ) dataset (Kwiatkowski et al. 2019). It converts
NQ's dataset into an AS2 (answer-sentence-selection) format.

The dataset details can be found in the paper at
https://arxiv.org/abs/1911.04118

The dataset can be downloaded at
https://d3t7erp6ge410c.cloudfront.net/tanda-aaai-2020/data/asnq.tar

"""


import csv
import os

import datasets


_CITATION = """\
@article{garg2019tanda,
    title={TANDA: Transfer and Adapt Pre-Trained Transformer Models for Answer Sentence Selection},
    author={Siddhant Garg and Thuy Vu and Alessandro Moschitti},
    year={2019},
    eprint={1911.04118},
}
"""

_DESCRIPTION = """\
ASNQ is a dataset for answer sentence selection derived from
Google's Natural Questions (NQ) dataset (Kwiatkowski et al. 2019).

Each example contains a question, candidate sentence, label indicating whether or not
the sentence answers the question, and two additional features --
sentence_in_long_answer and short_answer_in_sentence indicating whether ot not the
candidate sentence is contained in the long_answer and if the short_answer is in the candidate sentence.

For more details please see
https://arxiv.org/pdf/1911.04118.pdf

and

https://research.google/pubs/pub47761/
"""

_URL = "data/asnq.zip"


class ASNQ(datasets.GeneratorBasedBuilder):
    """ASNQ is a dataset for answer sentence selection derived
    ASNQ is a dataset for answer sentence selection derived from
    Google's Natural Questions (NQ) dataset (Kwiatkowski et al. 2019).

    The dataset details can be found in the paper:
    https://arxiv.org/abs/1911.04118
    """

    VERSION = datasets.Version("1.0.0")

    def _info(self):

        return datasets.DatasetInfo(
            # This is the description that will appear on the datasets page.
            description=_DESCRIPTION,
            # This defines the different columns of the dataset and their types
            features=datasets.Features(
                {
                    "question": datasets.Value("string"),
                    "sentence": datasets.Value("string"),
                    "label": datasets.ClassLabel(names=["neg", "pos"]),
                    "sentence_in_long_answer": datasets.Value("bool"),
                    "short_answer_in_sentence": datasets.Value("bool"),
                }
            ),
            # No default supervised_keys
            supervised_keys=None,
            # Homepage of the dataset for documentation
            homepage="https://github.com/alexa/wqa_tanda#answer-sentence-natural-questions-asnq",
            citation=_CITATION,
        )

    def _split_generators(self, dl_manager):
        """Returns SplitGenerators."""
        # dl_manager is a datasets.download.DownloadManager that can be used to
        # download and extract URLs
        dl_dir = dl_manager.download_and_extract(_URL)
        data_dir = os.path.join(dl_dir, "data", "asnq")
        return [
            datasets.SplitGenerator(
                name=datasets.Split.TRAIN,
                # These kwargs will be passed to _generate_examples
                gen_kwargs={
                    "filepath": os.path.join(data_dir, "train.tsv"),
                    "split": "train",
                },
            ),
            datasets.SplitGenerator(
                name=datasets.Split.VALIDATION,
                # These kwargs will be passed to _generate_examples
                gen_kwargs={
                    "filepath": os.path.join(data_dir, "dev.tsv"),
                    "split": "dev",
                },
            ),
        ]

    def _generate_examples(self, filepath, split):
        """Yields examples.

        Original dataset contains labels '1', '2', '3' and '4', with labels
        '1', '2' and '3' considered negative (sentence does not answer the question),
        and label '4' considered positive (sentence does answer the question).
        We map these labels to two classes, returning the other properties as additional
        features."""

        # Mapping of dataset's original labels to a tuple of
        # (label, sentence_in_long_answer, short_answer_in_sentence)
        label_map = {
            "1": ("neg", False, False),
            "2": ("neg", False, True),
            "3": ("neg", True, False),
            "4": ("pos", True, True),
        }
        with open(filepath, encoding="utf-8") as tsvfile:
            tsvreader = csv.reader(tsvfile, delimiter="\t")
            for id_, row in enumerate(tsvreader):
                question, sentence, orig_label = row
                label, sentence_in_long_answer, short_answer_in_sentence = label_map[orig_label]
                yield id_, {
                    "question": question,
                    "sentence": sentence,
                    "label": label,
                    "sentence_in_long_answer": sentence_in_long_answer,
                    "short_answer_in_sentence": short_answer_in_sentence,
                }