Hansel.py

# coding=utf-8

"""Hansel: A Chinese Few-Shot and Zero-Shot Entity Linking Benchmark"""


import json
import os

import datasets



_HANSEL_CITATION = """\
@misc{xu2022hansel,
  title = {Hansel: A Chinese Few-Shot and Zero-Shot Entity Linking Benchmark},
  author = {Xu, Zhenran and Shan, Zifei and Li, Yuxin and Hu, Baotian and Qin, Bing},
  publisher = {arXiv},
  year = {2022},
  url = {https://arxiv.org/abs/2207.13005}
}

"""

_HANSEL_DESCRIPTION = """\
Hansel is a high-quality human-annotated Chinese entity linking (EL) dataset, used for testing Chinese EL systems' generalization ability to tail entities and emerging entities.
The test set contains Few-shot (FS) and zero-shot (ZS) slices, has 10K examples and uses Wikidata as the corresponding knowledge base.
The training and validation sets are from Wikipedia hyperlinks, useful for large-scale pretraining of Chinese EL systems.

"""


_URLS = {
    "train": "hansel-train.jsonl",
    "val": "hansel-val.jsonl",
    "hansel-fs": "hansel-few-shot-v1.jsonl",
    "hansel-zs": "hansel-zero-shot-v1.jsonl",
}

logger = datasets.logging.get_logger(__name__)


class HanselConfig(datasets.BuilderConfig):
    """BuilderConfig for HanselConfig."""

    def __init__(self, features, data_url, citation, url, **kwargs):
        """BuilderConfig for Hansel.

        Args:
          features: `list[string]`, list of the features that will appear in the
            feature dict. Should not include "label".
          data_url: `string`, url to download the zip file from.
          citation: `string`, citation for the data set.
          url: `string`, url for information about the data set.
          **kwargs: keyword arguments forwarded to super.
        """
        super(HanselConfig, self).__init__(version=datasets.Version("1.0.0"), **kwargs)
        self.features = features
        self.data_url = data_url
        self.citation = citation
        self.url = url


class Hansel(datasets.GeneratorBasedBuilder):
    """The Hansel benchmark."""

    BUILDER_CONFIGS = [
        HanselConfig(
            name="wiki",
            description=_HANSEL_DESCRIPTION,
            features=["id", "text", "start", "end", "mention", "gold_id"],
            data_url="https://huggingface.co./datasets/HIT-TMG/Hansel/blob/main/",
            citation=_HANSEL_CITATION,
            url="https://github.com/HITsz-TMG/Hansel",
        )
        HanselConfig(
            name="hansel-few-shot",
            description=_HANSEL_DESCRIPTION,
            features=["id", "text", "start", "end", "mention", "gold_id", "source", "domain"],
            data_url="https://huggingface.co./datasets/HIT-TMG/Hansel/blob/main/",
            citation=_HANSEL_CITATION,
            url="https://github.com/HITsz-TMG/Hansel",
        )
        HanselConfig(
            name="hansel-zero-shot",
            description=_HANSEL_DESCRIPTION,
            features=["id", "text", "start", "end", "mention", "gold_id", "source", "domain"],
            data_url="https://huggingface.co./datasets/HIT-TMG/Hansel/blob/main/",
            citation=_HANSEL_CITATION,
            url="https://github.com/HITsz-TMG/Hansel",
        )
    ]

    def _info(self):
        features = {feature: datasets.Value("string") for feature in self.config.features}
        features["start"] = datasets.Value("int64")
        features["end"] = datasets.Value("int64")

        return datasets.DatasetInfo(
            description=self.config.description,
            features=datasets.Features(features),
            homepage=self.config.url,
            citation=self.config.citation
        )


    def _split_generators(self, dl_manager):

        urls_to_download = self._URLS
        downloaded_files = dl_manager.download_and_extract(urls_to_download)

        if "hansel-few" in self.config.name:
            return [
                datasets.SplitGenerator(
                    name=datasets.Split.TEST,
                    gen_kwargs={
                        "data_file": downloaded_files["hansel-fs"]),
                        "split": datasets.Split.TEST,
                    },
                ),
            ]
        if "hansel-zero" in self.config.name:
            return [
                datasets.SplitGenerator(
                    name=datasets.Split.TEST,
                    gen_kwargs={
                        "data_file": downloaded_files["hansel-zs"],
                        "split": datasets.Split.TEST,
                    },
                ),
            ]
        return [
            datasets.SplitGenerator(
                name=datasets.Split.TRAIN,
                gen_kwargs={
                    "data_file": downloaded_files["train"],
                    "split": datasets.Split.TRAIN,
                },
            ),
            datasets.SplitGenerator(
                name=datasets.Split.VALIDATION,
                gen_kwargs={
                    "data_file": downloaded_files["val"],
                    "split": datasets.Split.VALIDATION,
                },
            ),
        ]

    def _generate_examples(self, data_file, split):
        logger.info("generating examples from = %s", data_file)
        with open(data_file, encoding="utf-8") as f:
            for line in f:
                temDict = json.loads(line)
                key = temDict["id"]
                if "hansel" in self.config.name:
                    yield key, {
                                "text": temDict["text"],
                                "start": temDict["start"],
                                "end": temDict["end"],
                                "mention": temDict["mention"],
                                "gold_id": temDict["gold_id"],
                                "source": temDict["source"],
                                "domain": temDict["domain"],
                            }
                else:
                    yield key, {
                                "text": temDict["text"],
                                "start": temDict["start"],
                                "end": temDict["end"],
                                "mention": temDict["mention"],
                                "gold_id": temDict["gold_id"],
                            }