HRLB^2: A reinforcement learning based framework for believable bots Academic Article in Scopus