Semantic Properties of cosine based bias scores for word embeddings

Open in new window