The idea is to build a predictive system that predicts the degree of likeness of new food joints for users based on users past historical data on a global geographical scale.
e.g. If user travels to California from Boston, the system will predict which restaurants he will like in his five-mile radius and how will he rate the restaurant based on his/her previous pattern on Yelp.