This document discusses using latent Dirichlet allocation (LDA) to classify topics in restaurant reviews and recommend restaurants. It finds that LDA reduces the feature dimension by over 99% and computation time by over 90% compared to bag-of-words modeling. LDA identifies 12 topics from the reviews, including Japanese, Mexican, brunch, bar atmosphere, pizza, Indian, Asian, fast food, sweets, BBQ, and a topic on bad service. The document evaluates recommendation accuracy using a normalized distance-based performance measure to minimize the difference between recommended and actual rankings.