Situation Awareness for Proactive In-Car Recommendations of Points-Of-Interest (POI)

Recommender systems have become widely used for providing personalized product, media and travel information. Our goal is to apply them in automotive scenarios by recommending Points-Of-Interest (POI) such as fuel stations, restaurants or parking places. Furthermore, driver distraction caused by user interaction is a problem in automotive scenarios. We aim to reduce interaction by providing recommendations in a proactive manner. To give an example, proactive recommendations of fuel stations are of interest for the driver if he is expected to run out of fuel or reachable fuel stations. To enable proactive in time recommendations, context plays a fundamental role. It describes user situations which determine the relevance of recommendations. Our contribution is a new model for situation awareness tailored to proactive recommendations in automotive scenarios. Unlike other approaches, our model allows for incorporating past and future situations for assessing the relevance of a recommendation in the present. The presented model is based on fuzzy logic to cope with different sources of uncertainty. The implemented scenario is a proactive POI recommender for fuel stations using situation descriptions for fuel level states and number of reachable fuel stations. The evaluation shows that the model allows to determine significant situation changes. This information can then be used to derive decisions on when to recommend a fuel station.