Driver Recommendations of POIs using a Semantic Content-based Approach

In this paper, we present a semantic content-based approach that is employed to study driver preferences for Points of Interest (POIs), e.g. banks, grocery stores, etc., and provide recommendations for new POIs. Initially, logs about the places that the driver visits are collected from the cloud-connected navigation application running in the car. Data about the visited places is gathered from multiple sources and represented semantically in RDF by 'lifting' it. This semantic data is then combined with driver context and then input into a machine learning algorithm that produces a probabilistic model of the driver's preferences of POIs. When the driver searches for POIs in an unknown area, this preference model is used to recommend places that he is most likely to prefer using a nearest-neighbor approach. In this paper, we describe the details of this content-based approach for recommendation, along with the results of a user study that was conducted to evaluate the approach.