k-GAgent: Negotiating Agents Considering Interdependencies Between Issues

Abstract

Bilateral multi-issue closed negotiation is an important class for real-life negotiations. Usually, negotiation problems have constraints such as a complex and unknown opponent’s utility in real time, or time discounting. Recently, the attention of this study has focused on the nonlinear utility functions. In nonlinear utility functions, most of the negotiation strategies for linear utility functions can’t adopt to the scenarios of nonlinear utility functions. In this chapter, we propose the estimating method for the pareto frontier based on the opponent’s bids. In the proposed method, the opponent’s bid is divided into small elements considering the combinations between issues, and counted the number of the opponent’s proposing to estimated the opponent’s utility function. In addition, the genetic algorithm is employed to the proposed method to search the pareto optimal bids based on the opponent’s estimated utility and own utility. Experimental results demonstrate that our proposed method considering the interdependency between issues can search the pareto optimal bids.