Word embeddings, or vector representations of words, are commonly used in computer science to work with and analyze text. They are particularly useful as a powerful off-the-shelf tool when using open-source word embeddings previously generated by Google, Facebook, or other technology companies based on web crawls. We present the background and justifications for using vectors to represent meaning, how word embeddings are created, and applications of word embeddings in the social sciences and humanities. We’ll also touch on algorithmic bias, how this is presented in word embeddings, and what practitioners should be aware of.

LEVEL: Beginner
NOTES: none