Kieraj Mumick, Igor Pogorelskiy, Richard Kitain, Alec Olesky
WordRank is a project that has the goal of determining the reliability of ratings and reviews left by users on Yelp. The way in which this reliability would be determined would be based on an algorithm that we create that predicts what the rating of a specific review would be. The way in which how reliable a review is would be determined by how far away that rating for that review is as compared to what our predicted rating. If the rating for a specific review is close to what our predicted rating is, that would mean that the user who left the review left an accurate rating based on the review he left, otherwise he may have left a pretty inaccurate one. The motivation behind this idea comes from the fact that there are often many cases in which the actual rating does not match what the review says - the number that a person assigns to the quality of a restaurant or business is not close to what their review says it should be. There are many reasons for why this could be the case. One reason could be that each user has a different idea of what an x-star rating means (where x is some integer from 1 to 5, inclusive), and as a result a giving a location 4-stars may mean the same thing to one person as giving a location 2-starts may mean to a different person. This algorithm would be able to detect this and make the overall ratings more consistent across all of the reviews. By making these ratings more consistent across all the reviews, we are able to more accurately portray the real quality of each restaurant/business (as opposed to each restaurant/business nominal quality, as determined by the actual ratings given by the user).These real values will more accurately represent to other Yelp users the quality of the restaurants/businesses that they would like to go to. Through this, we are able to determine the reliability of rating and reviews left by users on Yelp.