Inferring ratings for custom trips from rich GPS traces

Trip planning services are employed extensively by users to compute paths between locations and navigate within a road network. In some real-world scenarios such as planning for a hiking trip or running training, users usually require personalized trip planning. Although some existing systems can recommend trips that other users have posted, along with a set of ratings w.r.t. the difficulty of the route, conditions, or the enjoyment it provides. Very often though users want to define a custom trip that fits their personal needs, for which existing systems are unable to provide any rating. In this paper we therefore define the problem of inferring ratings for custom trips. We also outline a solution to infer ratings by utilizing the ratings of trips previously posted by users and their similarity with a given custom trip. Finally, we present the results of preliminary experiments were we evaluate the efficiency of our proposed approach on inferring ratings for trips related to hiking and other similar activities.