This document describes Yelper Helper, a personalized review engine for Yelp users. It aims to determine the usefulness of new reviews and compute user similarity. It does this by using a data structure containing over 363,000 reviews from over 112,000 users of 3,536 businesses. It predicts the number of "useful" tags for a review using a zero-inflated Poisson regression model after performing feature selection and model selection. User similarity is computed through a similarity matrix calculating Euclidean distance between user-taste and restaurant-category matrices generated through collaborative filtering of user ratings data.