Language corpora for the Dutch medical domain

TL;DR

This paper introduces the first large-scale Dutch medical language corpus, consisting of approximately 35 billion tokens from 100 million documents, to advance NLP in the medical domain.

Contribution

It creates and publicly releases a comprehensive Dutch medical corpus by translating datasets, identifying medical texts, and extracting resources, enabling improved NLP applications.

Findings

Corpus contains ~35 billion tokens across 100 million documents.

The corpus is freely available on Hugging Face.

Establishes a foundational resource for Dutch medical NLP.

Abstract

\textbf{Background:} Dutch medical corpora are scarce, limiting NLP development. \\ \textbf{Methods:} We translated English datasets, identified medical text in generic corpora, and extracted open Dutch medical resources. \\ \textbf{Results:} The resulting corpus comprises $\pm$ 35 billion tokens across the medical domain in about 100 million documents, freely available on Hugging Face. \\ \textbf{Conclusion:} This work establishes the first large-scale Dutch medical language corpus for pre-training and downstream NLP tasks.