This dataset contains Turkish-language medical articles scraped from 14 official hospital and healthcare provider websites in Turkey. Each file corresponds to one source and is stored in efficient .parquet format.

It is designed for training and evaluating Turkish NLP models in the medical domain, including large language models (LLMs), health chatbots, summarizers, and classifiers.

🧾 Total articles: ~ 25,000
📦 Total file size: ~95MB
🗂️ Format: .parquet
👤 Curated by: umutertugrul
🔓 License: CC BY 4.0

🏷️ Included Sources

Filename	Source Institution
`acibadem.parquet`	Acıbadem Hospital
`anadolusaglik.parquet`	Anadolu Sağlık
`atlas.parquet`	Atlas Hospital
`baskentistanbul.parquet`	Başkent İstanbul
`bayindir.parquet`	Bayındır Hospital
`florence.parquet`	Florence Nightingale
`guven.parquet`	Güven Hospital
`liv.parquet`	Liv Hospital
`medicalpark.parquet`	Medical Park
`medicalpoint.parquet`	Medical Point
`medicana.parquet`	Medicana Hospitals
`medipol.parquet`	Medipol University Hospital
`memorial.parquet`	Memorial Hospital
`yeditepe.parquet`	Yeditepe University

Each file contains articles specific to that institution, covering a wide range of medical topics and specialties.

🔤 Schema

Each .parquet file may contain the following fields:

url: Original article URL
title: Title of the article
headings: Section headings or subtopics (if available)
text: Full body content of the article
publish_date: Date the article was originally published
update_date: Last updated date (if available)
scrape_date: Date when the article was collected
__source: Hospital/source name (automatically added during conversion)

🚀 Example Usage

from datasets import load_dataset, Features, Value

files = {
    "acibadem": "acibadem.parquet",
    "anadolusaglik": "anadolusaglik.parquet",
    "atlas": "atlas.parquet",
    "baskentistanbul": "baskentistanbul.parquet",
    "bayindir": "bayindir.parquet",
    "florence": "florence.parquet",
    "guven": "guven.parquet",
    "liv": "liv.parquet",
    "medicalpark": "medicalpark.parquet",
    "medicalpoint": "medicalpoint.parquet",
    "medicana": "medicana.parquet",
    "medipol": "medipol.parquet",
    "memorial": "memorial.parquet",
    "yeditepe": "yeditepe.parquet"
}

features = Features({
    "url": Value("string"),
    "title": Value("string"),
    "headings": Value("string"),
    "text": Value("string"),
    "publish_date": Value("string"),
    "update_date": Value("string"),
    "scrape_date": Value("string"),
    "__source": Value("string") 
})

ds_healthcare_1 = load_dataset(
    "umutertugrul/turkish-hospital-medical-articles",
    data_files=files,
    features=features
)

Downloads last month: 3,199