Do You Need Embeddings Trained on a Massive Specialized Corpus for Your Clinical Natural Language Processing Task? | Publicación