Personalized itinerary recommendation with time constraints using GPS datasets

Planning a personalized itinerary for an unfamiliar region requires much effort to design desirable travel plans. With the rapid development of location-based social network (LBSN) services, data mining techniques are utilized to retrieve useful information such as geographical features and social relationships. In this paper, we propose a personalized itinerary recommendation with time constraints (pirT) framework for the LBSN by exploiting geographical features and social relationships to recommend a personalized itinerary that satisfies user preferences (i.e., travel behaviors). In pirT, we have designed a user-based collaborative filtering with time preference (UTP) to explore user preferences by considering the visiting time of locations which the users have visited in our framework. UTP allows a tourist to find users with similar travel behaviors to those of the service requester in the past and to recommend interesting locations in the itineraries that these similar users have traveled to before. Subsequently, given a beginning location and a destination with a time constraint specified by the tourist, we devise the top-k\(A^*\) search-based recommendations and re-ranking itinerary candidate algorithms to efficiently plan the top k personalized itineraries. In the planning process, we simultaneously take account of the visiting time of locations, the transit time between locations, and the order of visiting locations. We conducted our experiments on the Gowalla dataset and demonstrated the effectiveness of our pirT framework comparing it with the personalized trip recommendation (PTR) framework. The results show that our pirT framework is superior to the PTR framework.