Learning of virtual words utilized in negotiation process between agents

Agent-based simulations are expected to enable analysis of complex social p henomena. In such simulations, one of the important behaviors of the agents is negotiation. Throughout the negotiations, the agents can make complex i nteractions with each other. Therefore, the ability of agents to perform ne gotiation is important in simulations of artificial societies. In this pape r, we focus on price negotiations, in which the two sides have opposing int erests. In the conventional price negotiation model, the process consists o f an alternate succession of directly presented offers and counter-offers e xchanging the desired prices. As an extended price negotiation model, we in troduce virtual words to mimic the negotiation techniques of humans for ind irectly presenting: the desired price. The process of the proposed negotiat ion model consists of an alternate succession of offers of desired price an d counter-offers of a word. The words represent the degree of the agent's d emand. We propose agents with reinforcement learning who can acquire the ab ility to distinguish words and use them to negotiate. As a result, we will show that the virtual words became meaningful in the process of negotiation s between agents whose negotiating strategies are acquired by reinforcement leaning.