Towards Better Service Personalization: Reinforcement Learning with Guarantee of User Preference | IEEE Conference Publication | IEEE Xplore