Personalized Itinerary Recommendation via Expectation-Maximization

The personalized itinerary recommendation problem in selecting a subset of locations to visit from among a larger set while maximizing the benefit for the tourist. In this work, we propose an efficient deterministic method for the recommendation of personalized itineraries consisting of a sequence of Points of Interest (POIs) that maximizes the expected user satisfaction and adheres to user time constraints. Experimental results on a large number of synthetic and real-world datasets demonstrate the high performance of our framework.