City Recommender System based on a Latent Topic Model

摘要

Travel recommendation is a very challenging for the application of recommender systems. A travel recommender system helps travelers in decision making processes when they are planning a trip, including the choice of destinations. In this research, our goal is to build algorithm for recommending venues for a user when he visits a city. One of the challenges is the recommendation of venues in a city that the user has never been (no history activities in that city - the cold-start problem). We cope with this challenge by exploiting the location information of users ' history activities as well as users' behavior when they visit a city. Our experiment is conducted on the dataset of FourSquare, a location-based social network.