• word2vec and GloVe are both methods for static embeddings
  • word2vec learns from within-document word associations based on physical proximity
  • GloVe learns from across-document word counts (think fancy, nonlinear TF-IDF)