Enhancing Inference in Relational Reinforcement Learning Via Truth Maintenance Systems | IEEE Conference Publication | IEEE Xplore