I recommend you read the paper here: http://nlp.stanford.edu/projects/glove/ http://nlp.stanford.edu/projects/glove/ It gives intuition/light math why the ana

I recommend you read the paper here: http://nlp.stanford.edu/projects/glove/ http://nlp.stanford.edu/projects/glove/ It gives intuition/light math why the analogy as translation in embedding space should work.