Originally we wanted to find identify fake and spam users on Yelp to help improve the Yelp user space, but we weren't 100% sure on what criteria to use. We eventually did compile a list of users who we felt had the potential to be fake (few friends, few posts, very low or high ratings, and over 1.5 points off from the average review). But needed to narrow down the criteria, mainly the review text itself. We decided to look at what words were frequently used for a specific business and tried to see if we can match generic words with the review. In the end, we we built a web app that would select a random business, get its word cloud, and compare the word cloud to a user on our fake list and see what the results are. The result is Yelpipedia: the dark side of Yelp.