Hybrid Q-Learning Applied to Ubiquitous recommender system

摘要

Ubiquitous information access becomes more and more important nowadays and research is aimed at making it adapted to users. Our work consists in applying machine learning techniques in order to bring a solution to some of the problems concerning the acceptance of the system by users. To achieve this, we propose a fundamental shift in terms of how we model the learning of recommender system: inspired by models of human reasoning developed in robotic, we combine reinforcement learning and case-base reasoning to define a recommendation process that uses these two approaches for generating recommendations on different context dimensions (social, temporal, geographic). We describe an implementation of the recommender system based on this framework. We also present preliminary results from experiments with the system and show how our approach increases the recommendation quality.