Kind of. LSI is a dimensionality-reduction method, which means it takes really sparse "bag of words" vectors and compresses them in a way which approximates the

Kind of. LSI is a dimensionality-reduction method, which means it takes really sparse "bag of words" vectors and compresses them in a way which approximates the original structure. Word2vec gives each word a non sparse vector based on the hidden units of a neural network language model. The model is trained to predict the middle word from it's surroundings, which makes it learn how words are used in context.