Mobile Recommendation within a Strong Privacy Oriented Paradigm - Naver Labs Europe
loader image


A new shift toward more user data privacy supported by recent regulations may deeply impact the way search and recommendation systems work. We explore in this paper some constraints and challenges raised by such paradigm shift in a live experiment aiming to develop an ambient and context aware personal recommender system for mobile users in a strong privacy by-design environment. We discuss the impact of a lack of precision for stay area detection and focus more specifically on semantic disambiguation challenges for point of interest inference to build a valuable user profile. We present and discuss our findings from a live experiment made in the cities of Lyon and Grenoble in France