Efficient profiling and distributed synchronization of statistical user models for content-targeting and real-time applications

This work presents an efficient approach for managing distributed user profiles in the use-case of the semi-autonomous or fully autonomous connected car. A user profile is a collection about user´s personal preferences and statistical data on user´s previous interactions with all compatible connected systems outside and inside the car. Managing the user profile practically means dynamically updating the profile based on the interaction with the vehicle´s human-machine interfaces and using it to answer complex queries placed by the vehicle´s applications. The queries can range between user interface personalization queries (what is the user´s favorite radio station), over product-placement related queries (are we going to pass next to a restaurant the user might be interested in) to safety-relevant queries (is the senior driver paying adequate attention to the direction of a detected obstacle). The focus is placed on reducing the computational complexity of user profiling and synchronization of distributed user profiles.