PHR: A Personalized Hidden Route Recommendation System Based on Hidden Markov Model

Abstract

Route recommendation based on users’ historical trajectories and behavior preferences is one of the important research problems. However, most of the existing work recommends a route based on the similarity among the routes in historical trajectories. As a result, hidden routes that also meet the users’ requirements cannot be explored. To solve this problem, we developed a system PHR that can recommend hidden routes to users employing the Hidden Markov Model, where a route recommendation problem is transformed to a point-of-interested (POI) sequence prediction. The system can return the top-k results including both explicit and hidden routes considering the personalized category sequence, route length, POI popularity, and visiting probabilities. The real check-in data from Foursquare is employed in this demo. The research can be used for travel itinerary plan or routine trip plan.