tacred.py

import json
import os

import datasets

_CITATION = """\
@inproceedings{zhang-etal-2017-position,
    title = "Position-aware Attention and Supervised Data Improve Slot Filling",
    author = "Zhang, Yuhao  and
      Zhong, Victor  and
      Chen, Danqi  and
      Angeli, Gabor  and
      Manning, Christopher D.",
    booktitle = "Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing",
    month = sep,
    year = "2017",
    address = "Copenhagen, Denmark",
    publisher = "Association for Computational Linguistics",
    url = "https://www.aclweb.org/anthology/D17-1004",
    doi = "10.18653/v1/D17-1004",
    pages = "35--45",
}

@inproceedings{alt-etal-2020-tacred,
    title = "{TACRED} Revisited: A Thorough Evaluation of the {TACRED} Relation Extraction Task",
    author = "Alt, Christoph  and
      Gabryszak, Aleksandra  and
      Hennig, Leonhard",
    booktitle = "Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics",
    month = jul,
    year = "2020",
    address = "Online",
    publisher = "Association for Computational Linguistics",
    url = "https://www.aclweb.org/anthology/2020.acl-main.142",
    doi = "10.18653/v1/2020.acl-main.142",
    pages = "1558--1569",
}
"""

# TODO: Add description of the dataset here
# You can copy an official description
_DESCRIPTION = """\
TACRED is a large-scale relation extraction dataset with 106,264 examples built over newswire
 and web text from the corpus used in the yearly TAC Knowledge Base Population (TAC KBP) challenges.
 Examples in TACRED cover 41 relation types as used in the TAC KBP challenges (e.g., per:schools_attended
 and org:members) or are labeled as no_relation if no defined relation is held. These examples are created
 by combining available human annotations from the TAC KBP challenges and crowdsourcing.

 Please see our EMNLP paper, or our EMNLP slides for full details.

Note: There is currently a label-corrected version of the TACRED dataset, which you should consider using instead of
the original version released in 2017. For more details on this new version, see the TACRED Revisited paper
published at ACL 2020.
"""

# TODO: Add a link to an official homepage for the dataset here
_HOMEPAGE = "https://nlp.stanford.edu/projects/tacred/"

# TODO: Add the licence for the dataset here if you can find it
_LICENSE = "LDC"

_URL = "https://catalog.ldc.upenn.edu/LDC2018T24"

# The HuggingFace dataset library don't host the datasets but only point to the original files
# This can be an arbitrary nested dict/list of URLs (see below in `_split_generators` method)
_PATCH_URLs = {
    "dev": "https://raw.githubusercontent.com/DFKI-NLP/tacrev/master/patch/dev_patch.json",
    "test": "https://raw.githubusercontent.com/DFKI-NLP/tacrev/master/patch/test_patch.json",
}

_CLASS_LABELS = [
    "no_relation",
    "org:alternate_names",
    "org:city_of_headquarters",
    "org:country_of_headquarters",
    "org:dissolved",
    "org:founded",
    "org:founded_by",
    "org:member_of",
    "org:members",
    "org:number_of_employees/members",
    "org:parents",
    "org:political/religious_affiliation",
    "org:shareholders",
    "org:stateorprovince_of_headquarters",
    "org:subsidiaries",
    "org:top_members/employees",
    "org:website",
    "per:age",
    "per:alternate_names",
    "per:cause_of_death",
    "per:charges",
    "per:children",
    "per:cities_of_residence",
    "per:city_of_birth",
    "per:city_of_death",
    "per:countries_of_residence",
    "per:country_of_birth",
    "per:country_of_death",
    "per:date_of_birth",
    "per:date_of_death",
    "per:employee_of",
    "per:origin",
    "per:other_family",
    "per:parents",
    "per:religion",
    "per:schools_attended",
    "per:siblings",
    "per:spouse",
    "per:stateorprovince_of_birth",
    "per:stateorprovince_of_death",
    "per:stateorprovinces_of_residence",
    "per:title",
]


def convert_ptb_token(token: str) -> str:
    """Convert PTB tokens to normal tokens"""
    return {
        "-lrb-": "(",
        "-rrb-": ")",
        "-lsb-": "[",
        "-rsb-": "]",
        "-lcb-": "{",
        "-rcb-": "}",
    }.get(token.lower(), token)


class Tacred(datasets.GeneratorBasedBuilder):
    """TACRED is a large-scale relation extraction dataset with 106,264 examples built over newswire
    and web text from the corpus used in the yearly TAC Knowledge Base Population (TAC KBP) challenges."""

    BUILDER_CONFIGS = [
        datasets.BuilderConfig(
            name="original", version=datasets.Version("1.0.0"), description="The original TACRED."
        ),
        datasets.BuilderConfig(
            name="revised",
            version=datasets.Version("1.0.0"),
            description="The revised TACRED (corrected labels in dev and test split).",
        ),
    ]

    DEFAULT_CONFIG_NAME = "original"  # type: ignore

    @property
    def manual_download_instructions(self):
        return (
            "To use TACRED you have to download it manually. "
            "It is available via the LDC at https://catalog.ldc.upenn.edu/LDC2018T24"
            "Please extract all files in one folder and load the dataset with: "
            "`datasets.load_dataset('tacred', data_dir='path/to/folder/folder_name')`"
        )

    def _info(self):
        features = datasets.Features(
            {
                "id": datasets.Value("string"),
                "docid": datasets.Value("string"),
                "tokens": datasets.Sequence(datasets.Value("string")),
                "subj_start": datasets.Value("int32"),
                "subj_end": datasets.Value("int32"),
                "subj_type": datasets.Value("string"),
                "obj_start": datasets.Value("int32"),
                "obj_end": datasets.Value("int32"),
                "obj_type": datasets.Value("string"),
                "pos_tags": datasets.Sequence(datasets.Value("string")),
                "ner_tags": datasets.Sequence(datasets.Value("string")),
                "stanford_deprel": datasets.Sequence(datasets.Value("string")),
                "stanford_head": datasets.Sequence(datasets.Value("int32")),
                "relation": datasets.ClassLabel(names=_CLASS_LABELS),
            }
        )

        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=features,  # Here we define them above because they are different between the two configurations
            # 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=_HOMEPAGE,
            # License for the dataset if available
            license=_LICENSE,
            # Citation for the dataset
            citation=_CITATION,
        )

    def _split_generators(self, dl_manager):
        """Returns SplitGenerators."""
        # If several configurations are possible (listed in BUILDER_CONFIGS), the configuration selected by the user is in self.config.name

        # dl_manager is a datasets.download.DownloadManager that can be used to download and extract URLs
        # It can accept any type or nested list/dict and will give back the same structure with the url replaced with path to local files.
        # By default the archives will be extracted and a path to a cached folder where they are extracted is returned instead of the archive
        patch_files = {}
        if self.config.name == "revised":
            patch_files = dl_manager.download_and_extract(_PATCH_URLs)

        data_dir = os.path.abspath(os.path.expanduser(dl_manager.manual_dir))

        if not os.path.exists(data_dir):
            raise FileNotFoundError(
                "{} does not exist. Make sure you insert a manual dir via `datasets.load_dataset('tacred', data_dir=...)` that includes the unzipped files from the TACRED_LDC zip. Manual download instructions: {}".format(
                    data_dir, self.manual_download_instructions
                )
            )

        return [
            datasets.SplitGenerator(
                name=datasets.Split.TRAIN,
                gen_kwargs={
                    "filepath": os.path.join(data_dir, "train.json"),
                    "patch_filepath": None,
                },
            ),
            datasets.SplitGenerator(
                name=datasets.Split.TEST,
                gen_kwargs={
                    "filepath": os.path.join(data_dir, "test.json"),
                    "patch_filepath": patch_files.get("test"),
                },
            ),
            datasets.SplitGenerator(
                name=datasets.Split.VALIDATION,
                gen_kwargs={
                    "filepath": os.path.join(data_dir, "dev.json"),
                    "patch_filepath": patch_files.get("dev"),
                },
            ),
        ]

    def _generate_examples(self, filepath, patch_filepath):
        """Yields examples."""
        # TODO: This method will receive as arguments the `gen_kwargs` defined in the previous `_split_generators` method.
        # It is in charge of opening the given file and yielding (key, example) tuples from the dataset
        # The key is not important, it's more here for legacy reason (legacy from tfds)
        patch_examples = {}
        if patch_filepath is not None:
            with open(patch_filepath, encoding="utf-8") as f:
                patch_examples = {example["id"]: example for example in json.load(f)}

        with open(filepath, encoding="utf-8") as f:
            data = json.load(f)
            for example in data:
                id_ = example["id"]

                if id_ in patch_examples:
                    example.update(patch_examples[id_])

                yield id_, {
                    "id": example["id"],
                    "docid": example["docid"],
                    "tokens": [convert_ptb_token(token) for token in example["token"]],
                    "subj_start": example["subj_start"],
                    "subj_end": example["subj_end"] + 1,  # make end offset exclusive
                    "subj_type": example["subj_type"],
                    "obj_start": example["obj_start"],
                    "obj_end": example["obj_end"] + 1,  # make end offset exclusive
                    "obj_type": example["obj_type"],
                    "relation": example["relation"],
                    "pos_tags": example["stanford_pos"],
                    "ner_tags": example["stanford_ner"],
                    "stanford_deprel": example["stanford_deprel"],
                    "stanford_head": [head - 1 for head in example["stanford_head"]]  # make offsets 0-based
                }