Surveys without Questions: A Reinforcement Learning Approach