Medical FineWeb Dataset Card
Overview
Medical FineWeb is a curated subset of medical content from the HuggingFace FineWeb dataset. It is designed for medical NLP applications, featuring high-quality, filtered, and annotated web texts.
Key Features
- High-quality scoring (TF-IDF, semantic analysis)
- Medical keyword categorization
- Deduplication
- Scalable, multi-threaded processing
- Automatic metadata and annotation
Use Cases
- Language Modeling
- Text Classification
- Medical Information Extraction
- Question Answering
- Text Generation
Languages
- Primarily English
- Others as available in FineWeb
Data Format
Each record is a JSON object:
{
"text": "...",
"record_id": "...",
"url": "...",
"date": "...",
"language": "en",
"language_score": 0.95,
"token_count": 1547,
"medical_keyword_count": 42,
"quality_score": 0.87,
"content_hash": "...",
"processing_timestamp": 1705123456.789
}