Spanish Biomedical and Clinical Language Embeddings

TL;DR

This paper presents Spanish biomedical and clinical language embeddings created with FastText, demonstrating improved performance with increased data and using BPE for sub-word representations.

Contribution

It introduces Spanish biomedical embeddings using FastText and BPE, showing that larger datasets lead to better representations.

Findings

Biomedical embeddings outperform previous versions

More data improves embedding quality

BPE effectively captures sub-word information

Abstract

We computed both Word and Sub-word Embeddings using FastText. For Sub-word embeddings we selected Byte Pair Encoding (BPE) algorithm to represent the sub-words. We evaluated the Biomedical Word Embeddings obtaining better results than previous versions showing the implication that with more data, we obtain better representations.