A Negotiation Learning Model for Open Multi-Agent Environments

The paper presents a model of heuristic negotiation between self-interested agents which allows the use of arguments, negotiation over multiple issues of the negotiation object, single and multi-party negotiation, and learning of the agent's negotiation primitives. The model uses negotiation objects and negotiation frames to separate the object of negotiation from the negotiation process. In order to negotiate strategically, the agents use a reinforcement learning algorithm applied on a specific state space representation of the negotiation process.