Yelp random reviews

Titipat Ahakulvisut, Zaw Htet Aung

Abstract

Reviews by normal people are an important part of the processes of deciding which restaurant to go to, doctor, or other general establishment. However, it is hard to understand how review ratings and mood affect the language of a review. In this work, we developed a new technique that can generate high quality reviews using as inputs review stars, business type, coolness, usefulness, and funny. Our technique is based on a continuous review sentence model using word vectors. We use regression model to predict potential vector of the next sentence then use nearest neighbor with randomness to pick the next sentence. We developed a website and free-software that can generate an unbounded number of reviews on the fly based on user inputs. Our work promises to help Yelp understand how features of the review related to their language, how review can be randomly generated which help understand fake reviews or randomly generated review.