medical_wikipedia.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.
"""MedWiki."""

import os

import datasets
import pandas as pd


_CITATION = """\
@article{corbeil2024iryonlp,
  title={IryoNLP at MEDIQA-CORR 2024: Tackling the Medical Error Detection & Correction Task On the Shoulders of Medical Agents},
  author={Jean-Philippe Corbeil},
  journal={arXiv preprint arXiv:2404.15488},
  year={2024}
}
"""

_DESCRIPTION = """\
This is a filtered version of the `Cohere/wikipedia-22-12` on medical topic articles using `MaartenGr/BERTopic_Wikipedia`. Keep note that some articles in the viewer might seem off topic, but usually they are related in some way (e.g. World War I is linked to the Spanish Flu). This is artefacts of some noise in the topic modelling.
"""
_HOMEPAGE = ""
_LICENSE = "CC-BY-SA"
_URLS = {
    "first_domain": "Cohere/wikipedia-22-12",
}


class MedWikiDataset(datasets.GeneratorBasedBuilder):
    """Medical Wikipedia Articles."""

    VERSION = datasets.Version("0.0.1")

    BUILDER_CONFIGS = [
        datasets.BuilderConfig(name="first_domain", version=VERSION, description="This part of my dataset covers a first domain"),
    ]

    DEFAULT_CONFIG_NAME = "first_domain"

    def _info(self):
        features = datasets.Features(
            {
                "wiki_id": datasets.Value("int32"),
                "title": datasets.Value("string"),
                "text": datasets.Value("string"),
                "paragraph_id": datasets.Value("int32"),
            }
        )

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

    def _split_generators(self, dl_manager):
        urls = _URLS[self.config.name]
        dataset = datasets.load_dataset(urls, "en", trust_remote_code=True, cache_dir="/Users/jcorbeil/Documents/datasets/TEMP")
        path = dl_manager.download_and_extract("https://huggingface.co/datasets/jpcorb20/medical_wikipedia/resolve/main/med_topics.csv")
        df = pd.read_csv(path)
        med_wiki_ids = set(df["wiki_id"].values.tolist())
        return [
            datasets.SplitGenerator(
                name=datasets.Split.TRAIN,
                gen_kwargs={
                    "dataset": dataset["train"],
                    "med_wiki_ids": med_wiki_ids,
                },
            ),
        ]

    def _generate_examples(self, dataset, med_wiki_ids):
        count = -1
        for data in dataset:
            if data["wiki_id"] in med_wiki_ids:
                count += 1
                yield count, {
                    "wiki_id": data["wiki_id"],
                    "title": data["title"],
                    "text": data["text"],
                    "paragraph_id": data["paragraph_id"],
                }