We present a novel bilateral negotiation model that allows a self-interested\nagent to learn how to negotiate over multiple issues in the presence of user\npreference uncertainty. The model relies upon interpretable strategy templates\nrepresenting the tactics the agent should employ during the negotiation and\nlearns template parameters to maximize the average utility received over\nmultiple negotiations, thus resulting in optimal bid acceptance and generation.\nOur model also uses deep reinforcement learning to evaluate threshold utility\nvalues, for those tactics that require them, thereby deriving optimal utilities\nfor every environment state. To handle user preference uncertainty, the model\nrelies on a stochastic search to find user model that best agrees with a given\npartial preference profile. Multi-objective optimization and multi-criteria\ndecision-making methods are applied at negotiation time to generate\nPareto-optimal outcomes thereby increasing the number of successful (win-win)\nnegotiations. Rigorous experimental evaluations show that the agent employing\nour model outperforms the winning agents of the 10th Automated Negotiating\nAgents Competition (ANAC'19) in terms of individual as well as social-welfare\nutilities.\n