README.md

metadata

license: cc-by-4.0
language:
  - tr
task_categories:
  - text-classification
  - text-generation
  - summarization
  - question-answering
tags:
  - medical
size_categories:
  - 10K<n<100K
configs:
  - config_name: default
    data_files:
      - split: acibadem
        path: data/acibadem-*
      - split: anadolusaglik
        path: data/anadolusaglik-*
      - split: atlas
        path: data/atlas-*
      - split: baskentistanbul
        path: data/baskentistanbul-*
      - split: bayindir
        path: data/bayindir-*
      - split: florence
        path: data/florence-*
      - split: guven
        path: data/guven-*
      - split: liv
        path: data/liv-*
      - split: medicalpark
        path: data/medicalpark-*
      - split: medicalpoint
        path: data/medicalpoint-*
      - split: medicana
        path: data/medicana-*
      - split: medipol
        path: data/medipol-*
      - split: memorial
        path: data/memorial-*
      - split: yeditepe
        path: data/yeditepe-*
dataset_info:
  features:
    - name: url
      dtype: string
    - name: title
      dtype: string
    - name: text
      dtype: string
    - name: publish_date
      dtype: string
    - name: update_date
      dtype: string
    - name: scrape_date
      dtype: string
  splits:
    - name: acibadem
      num_bytes: 59871616
      num_examples: 6339
    - name: anadolusaglik
      num_bytes: 7711085
      num_examples: 1012
    - name: atlas
      num_bytes: 659835
      num_examples: 130
    - name: baskentistanbul
      num_bytes: 1632855
      num_examples: 394
    - name: bayindir
      num_bytes: 3985692
      num_examples: 690
    - name: florence
      num_bytes: 12815638
      num_examples: 1641
    - name: guven
      num_bytes: 4279987
      num_examples: 666
    - name: liv
      num_bytes: 19820891
      num_examples: 2836
    - name: medicalpark
      num_bytes: 3323029
      num_examples: 371
    - name: medicalpoint
      num_bytes: 2374940
      num_examples: 654
    - name: medicana
      num_bytes: 17090402
      num_examples: 2163
    - name: medipol
      num_bytes: 7876549
      num_examples: 1380
    - name: memorial
      num_bytes: 44864194
      num_examples: 5338
    - name: yeditepe
      num_bytes: 5203404
      num_examples: 998
  download_size: 89711227
  dataset_size: 191510117

🏥 Turkish Medical Articles from 14 Hospital Websites

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
)