1 code implementation • ACL 2020 • Kenneth Joseph, Jonathan H. Morgan
We find that biases found in word embeddings do, on average, closely mirror survey data across seventeen dimensions of social meaning.
Word Embeddings