A novel strategy for efficient negotiation in complex environments

Abstract

A complex and challenging bilateral negotiation environment for rational autonomous agents is where agents negotiate multi-issue contracts in unknown application domains against unknown opponents under real-time constraints. In this paper we present a novel negotiation strategy called EMAR for this kind of environment which is based on a combination of Empirical Mode Decomposition (EMD) and Autoregressive Moving Average (ARMA). EMAR enables a negotiating agent to adjust its target utility and concession rate adaptively in real-time according to the behavior of its opponent. The experimental results show that this new strategy outperforms the best agents from the latest Automated Negotiation Agents (ANAC) Competition in a wide range of application domains. © 2012 Springer-Verlag.