This is the output of a word2vec python processing on a 229K sample of Yelp reviews.They were processed with gensim, then the most common 1500 words were plotted with TSNE in 2 dimensions. The color shows the sentiment polarity (red-blue, neg-pos) based on simple word sentiment from AFINN-111.txt. As you can see, more positive words (blue) appears in the reviews.
Click a dot to show the word associated with it; click the word to hide it again. Ask me for more info at @arnicas.