medwiki.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.
"""MedWiki is a large-scale sentence dataset collected from Wikipedia with medical entity (UMLS) annotations. This dataset is intended for pretraining"""

import csv
import json
import os

import datasets

_CITATION = """\
@inproceedings{medwiki,
    title={Cross-Domain Data Integration for Named Entity Disambiguation in Biomedical Text}, 
    author={Maya Varma and Laurel Orr and Sen Wu and Megan Leszczynski and Xiao Ling and Christopher Ré},
    year={2021},
    booktitle={Findings of the Association for Computational Linguistics: EMNLP 2021}
}
"""

_DESCRIPTION = """\
MedWiki is a large-scale sentence dataset collected from Wikipedia with medical entity (UMLS) annotations. This dataset is intended for pretraining.
"""

_HOMEPAGE = ""

_LICENSE = ""

_URLs = {"medwiki_full": "medwiki_full.zip", "medwiki_hq": "medwiki_hq.zip"}


class MedWiki(datasets.GeneratorBasedBuilder):
    """MedWiki: A Large-Scale Sentence Dataset with Medical Entity (UMLS) Annotations"""

    VERSION = datasets.Version("1.1.0")
    BUILDER_CONFIGS = [
        datasets.BuilderConfig(
            name="medwiki", 
            version=VERSION, 
            description="MedWiki: A Large-Scale Sentence Dataset with Medical Entity (UMLS) Annotations"),
    ]
    BUILDER_CONFIGS = [
        datasets.BuilderConfig(name="medwiki_full", version=VERSION, description="MedWiki (Full): A Large-Scale Sentence Dataset with Medical Entity (UMLS) Annotations."),
        datasets.BuilderConfig(name="medwiki_hq", version=VERSION, description="MedWiki (HQ): A Large-Scale Sentence Dataset with Medical Entity (UMLS) Annotations. The HQ (high quality) subset of MedWiki includes a portion of the dataset with higher-quality entity annotations."),
    ]


    def _info(self):
        features = datasets.Features(
            {
                "mentions": datasets.Sequence(datasets.Value("string")),
                "entities": datasets.Sequence(datasets.Value("string")),
                "entity_titles": datasets.Sequence(datasets.Value("string")),
                "types": datasets.Sequence(datasets.Sequence(datasets.Value("string"))),
                "spans": datasets.Sequence(datasets.Sequence(datasets.Value("int32"))),
                "sentence": datasets.Value("string"),
                "sent_idx_unq": datasets.Value("int32"),
            }
        )
        return datasets.DatasetInfo(
            description=_DESCRIPTION,
            features=features,  
            supervised_keys=None,
            homepage=_HOMEPAGE,
            license=_LICENSE,
            citation=_CITATION,
        )

    def _split_generators(self, dl_manager):
        """Returns SplitGenerators."""
        my_urls = _URLs[self.config.name]
        data_dir = dl_manager.download_and_extract(my_urls)
        
        #Adjust filenames for medwiki_hq subset
        ext = ""
        if self.config.name == "medwiki_hq": ext = "_hq"
        
        #Load splits
        return [
            datasets.SplitGenerator(
                name=datasets.Split.TRAIN,
                gen_kwargs={
                    "filepath": os.path.join(data_dir, f"train{ext}.jsonl"),
                    "split": "train",
                },
            ),
            datasets.SplitGenerator(
                name=datasets.Split.TEST,
                gen_kwargs={
                    "filepath": os.path.join(data_dir, f"test{ext}.jsonl"),
                    "split": "test"
                },
            ),
            datasets.SplitGenerator(
                name=datasets.Split.VALIDATION,
                gen_kwargs={
                    "filepath": os.path.join(data_dir, f"dev{ext}.jsonl"),
                    "split": "dev",
                },
            ),
        ]

    def _generate_examples(self, filepath, split ):
        """ Yields examples as (key, example) tuples. """

        with open(filepath, encoding="utf-8") as f:
            for id_, row in enumerate(f):
                data = json.loads(row)
                yield id_, {
                    "mentions": data["mentions"],
                    "entities": data["entities"],
                    "entity_titles": data['entity_titles'],
                    "types": data["types"],
                    "spans": data["spans"],
                    "sentence": data["sentence"],
                    "sent_idx_unq": data["sent_idx_unq"],
                }