Algebraic Reinforcement Learning - Hypothesis Induction for Relational Reinforcement Learning Using Term Generalization.