cosimlex.py

"""CoSimLex is a resource for evaluating graded word similarity in context."""


import csv

import datasets


_CITATION = """\
@inproceedings{armendariz-etal-2020-cosimlex,
    title = "{C}o{S}im{L}ex: A Resource for Evaluating Graded Word Similarity in Context",
    author = "Armendariz, Carlos Santos  and
      Purver, Matthew  and
      Ul{\v{c}}ar, Matej  and
      Pollak, Senja  and
      Ljube{\v{s}}i{\'c}, Nikola  and
      Granroth-Wilding, Mark",
    booktitle = "Proceedings of the 12th Language Resources and Evaluation Conference",
    month = may,
    year = "2020",
    url = "https://aclanthology.org/2020.lrec-1.720",
    pages = "5878--5886"
}
"""

_DESCRIPTION = """\
The dataset contains human similarity ratings for pairs of words. The annotators were presented with contexts that 
contained both of the words in the pair and the dataset features two different contexts per pair. The words were 
sourced from the English, Croatian, Finnish and Slovenian versions of the original Simlex dataset.
"""

_HOMEPAGE = "http://hdl.handle.net/11356/1308"

_LICENSE = "GNU General Public Licence, version 3"

_URLS = {
    "en": "https://www.clarin.si/repository/xmlui/bitstream/handle/11356/1308/cosimlex_en.csv",
    "fi": "https://www.clarin.si/repository/xmlui/bitstream/handle/11356/1308/cosimlex_fi.csv",
    "hr": "https://www.clarin.si/repository/xmlui/bitstream/handle/11356/1308/cosimlex_hr.csv",
    "sl": "https://www.clarin.si/repository/xmlui/bitstream/handle/11356/1308/cosimlex_sl.csv"
}


class CoSimLex(datasets.GeneratorBasedBuilder):
    """CoSimLex is a resource for evaluating graded word similarity in context."""

    VERSION = datasets.Version("1.0.0")

    BUILDER_CONFIGS = [
        datasets.BuilderConfig(name="en", version=VERSION, description="The English subset."),
        datasets.BuilderConfig(name="fi", version=VERSION, description="The Finnish subset."),
        datasets.BuilderConfig(name="hr", version=VERSION, description="The Croatian subset."),
        datasets.BuilderConfig(name="sl", version=VERSION, description="The Slovenian subset."),
    ]

    def _info(self):
        features = datasets.Features(
            {
                "word1": datasets.Value("string"), "word2": datasets.Value("string"),
                "context1": datasets.Value("string"), "context2": datasets.Value("string"),
                "sim1": datasets.Value("float32"), "sim2": datasets.Value("float32"),
                "stdev1": datasets.Value("float32"), "stdev2": datasets.Value("float32"),
                "pvalue": datasets.Value("float32"),
                "word1_context1": datasets.Value("string"), "word2_context1": datasets.Value("string"),
                "word1_context2": datasets.Value("string"), "word2_context2": datasets.Value("string")
            }
        )

        return datasets.DatasetInfo(
            description=_DESCRIPTION,
            features=features,
            homepage=_HOMEPAGE,
            license=_LICENSE,
            citation=_CITATION,
        )

    def _split_generators(self, dl_manager):
        urls = _URLS[self.config.name]
        file_path = dl_manager.download_and_extract(urls)
        return [
            datasets.SplitGenerator(
                name=datasets.Split.TRAIN,
                gen_kwargs={"file_path": file_path}
            )
        ]

    def _generate_examples(self, file_path):
        with open(file_path, encoding="utf-8") as f:
            reader = csv.reader(f, delimiter="\t", quotechar='"')
            header = next(reader)

            for i, row in enumerate(reader):
                yield i, {attr: value for attr, value in zip(header, row)}