xlel_wd.py

# 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.
# TODO: Address all TODOs and remove all explanatory comments
"""TODO: Add a description here."""


import json

import datasets

_CITATION = """\
@article{pratapa-etal-2022-multilingual,
title = {Multilingual Event Linking to Wikidata},
author = {Pratapa, Adithya and Gupta, Rishubh and Mitamura, Teruko},
publisher = {arXiv},
year = {2022},
url = {https://arxiv.org/abs/2204.06535},
}
"""

_DESCRIPTION = """\
XLEL-WD is a multilingual event linking dataset. \
This dataset contains mention references from multilingual Wikipedia/Wikinews articles to event items in Wikidata. \
The text descriptions for Wikidata events are compiled from Wikipedia articles.
"""

_HOMEPAGE = "https://github.com/adithya7/xlel-wd"

_LICENSE = "CC-BY-4.0"

# TODO: Add link to the official dataset URLs here
# The HuggingFace Datasets library doesn't host the datasets but only points to the original files.
# This can be an arbitrary nested dict/list of URLs (see below in `_split_generators` method)
_URLS = {
    "wikipedia-zero-shot": {
        "train": "wikipedia.train.jsonl",
        "dev": "wikipedia-zero-shot.dev.jsonl",
        "test": "wikipedia-zero-shot.test.jsonl",
    },
    "wikinews-zero-shot": {"test": "wikinews-zero-shot.test.jsonl"},
    "wikinews-cross-domain": {"test": "wikinews-cross-domain.test.jsonl"},
}

_WIKIPEDIA_ZERO_SHOT_LANGS = [
    "af",
    "ar",
    "be",
    "bg",
    "bn",
    "ca",
    "cs",
    "da",
    "de",
    "el",
    "en",
    "es",
    "fa",
    "fi",
    "fr",
    "he",
    "hi",
    "hu",
    "id",
    "it",
    "ja",
    "ko",
    "ml",
    "mr",
    "ms",
    "nl",
    "no",
    "pl",
    "pt",
    "ro",
    "ru",
    "si",
    "sk",
    "sl",
    "sr",
    "sv",
    "sw",
    "ta",
    "te",
    "th",
    "tr",
    "uk",
    "vi",
    "zh",
]

_WIKINEWS_CROSS_DOMAIN_LANGS = [
    "ar",
    "bg",
    "ca",
    "cs",
    "de",
    "el",
    "en",
    "es",
    "fi",
    "fr",
    "he",
    "hu",
    "it",
    "ja",
    "ko",
    "nl",
    "no",
    "pl",
    "pt",
    "ro",
    "ru",
    "sr",
    "sv",
    "ta",
    "tr",
    "uk",
    "zh",
]

_WIKINEWS_ZERO_SHOT_LANGS = [
    "ar",
    "cs",
    "de",
    "en",
    "es",
    "fi",
    "fr",
    "it",
    "ja",
    "ko",
    "nl",
    "no",
    "pl",
    "pt",
    "ru",
    "sr",
    "sv",
    "ta",
    "tr",
    "uk",
    "zh",
]

_TASK_NAMES = []
_TASK_DESCRIPTIONS = []

# wikipedia based tasks
_TASK_NAMES += ["wikipedia-zero-shot"]
_TASK_DESCRIPTIONS += [
    "This task requires linking mentions from multilingual wiki to Wikidata events (zero-shot evaluation)"
]

for lang in _WIKIPEDIA_ZERO_SHOT_LANGS:
    _TASK_NAMES += [f"wikipedia-zero-shot.{lang}"]
    _TASK_DESCRIPTIONS += [
        f"This task requires linking mentions from {lang}wiki to Wikidata events (zero-shot evaluation)."
    ]

# wikinews based tasks (zero-shot)
_TASK_NAMES += ["wikinews-zero-shot"]
_TASK_DESCRIPTIONS += [
    "This task requires linking mentions from multilingual wikinews to Wikidata events (zero-shot evaluation)."
]
for lang in _WIKINEWS_ZERO_SHOT_LANGS:
    _TASK_NAMES += [f"wikinews-zero-shot.{lang}"]
    _TASK_DESCRIPTIONS += [
        f"This task requires linking mentions from {lang}wikinews to Wikidata events (zero-shot evaluation)."
    ]

# wikinews based tasks (cross-domain)
_TASK_NAMES += ["wikinews-cross-domain"]
_TASK_DESCRIPTIONS += [
    "This task requires linking mentions from multilingual wikinews to Wikidata events (cross-domain evaluation)."
]
for lang in _WIKINEWS_CROSS_DOMAIN_LANGS:
    _TASK_NAMES += [f"wikinews-cross-domain.{lang}"]
    _TASK_DESCRIPTIONS += [
        f"This task requires linking mentions from {lang}wikinews to Wikidata events (cross-domain evaluation)."
    ]


class XlelWdConfig(datasets.BuilderConfig):
    """BuilderConfig for XLEL-WD"""

    def __init__(self, features, citation, url, **kwargs) -> None:
        super(XlelWdConfig, self).__init__(version=datasets.Version("1.0.0"), **kwargs)
        self.features = features
        self.citation = citation
        self.url = url


class XlelWd(datasets.GeneratorBasedBuilder):
    """A dataset for multilingual linking of event mentions to Wikidata."""

    VERSION = datasets.Version("1.0.0")
    # the features different slightly for Wikipedia and Wikinews
    # Wikinews dataset also contains the article title and publication date
    BUILDER_CONFIGS = [
        XlelWdConfig(
            name=name,
            description=desc,
            features=["mention", "context_left", "context_right", "context_lang"],
            citation=_CITATION,
            url=_URLS[name.split(".")[0]],
        )
        if name.startswith("wikipedia")
        else XlelWdConfig(
            name=name,
            description=desc,
            features=[
                "mention",
                "context_left",
                "context_right",
                "context_lang",
                "context_title",
                "context_date",
            ],
            citation=_CITATION,
            url=_URLS[name.split(".")[0]],
        )
        for name, desc in zip(_TASK_NAMES, _TASK_DESCRIPTIONS)
    ]

    def _info(self):

        features = {
            feature: datasets.Value("string") for feature in self.config.features
        }
        features["label_id"] = datasets.Value("string")
        return datasets.DatasetInfo(
            description=_DESCRIPTION + self.config.description,
            features=datasets.Features(features),
            homepage=_HOMEPAGE,
            license=_LICENSE,
            citation=self.config.citation,
        )

    def _split_generators(self, dl_manager):

        urls = _URLS[self.config.name.split(".")[0]]
        downloaded_files = dl_manager.download_and_extract(urls)
        if self.config.name.startswith("wikipedia"):
            return [
                datasets.SplitGenerator(
                    name=datasets.Split.TRAIN,
                    gen_kwargs={
                        "filepath": downloaded_files["train"],
                        "split": "train",
                    },
                ),
                datasets.SplitGenerator(
                    name=datasets.Split.VALIDATION,
                    gen_kwargs={
                        "filepath": downloaded_files["dev"],
                        "split": "dev",
                    },
                ),
                datasets.SplitGenerator(
                    name=datasets.Split.TEST,
                    gen_kwargs={
                        "filepath": downloaded_files["test"],
                        "split": "test",
                    },
                ),
            ]
        elif self.config.name.startswith("wikinews"):
            return [
                datasets.SplitGenerator(
                    name=datasets.Split.TEST,
                    gen_kwargs={
                        "filepath": downloaded_files["test"],
                        "split": "test",
                    },
                ),
            ]

    def _generate_examples(self, filepath, split):

        task_domain, *task_langs = self.config.name.split(".")
        with open(filepath, encoding="utf-8") as f:
            for key, row in enumerate(f):
                data = json.loads(row)
                # generate mention references for the specified language
                # if no language is specific in the config, return all
                if len(task_langs) == 0 or task_langs[0] == data["context_lang"]:
                    if task_domain.startswith("wikipedia"):
                        yield key, {
                            "mention": data["mention"],
                            "context_left": data["context_left"],
                            "context_right": data["context_right"],
                            "context_lang": data["context_lang"],
                            "label_id": data["label_id"],
                        }
                    elif task_domain.startswith("wikinews"):
                        yield key, {
                            "mention": data["mention"],
                            "context_left": data["context_left"],
                            "context_right": data["context_right"],
                            "context_lang": data["context_lang"],
                            "context_title": data["context_title"],
                            "context_date": data["context_date"],
                            "label_id": data["label_id"],
                        }