Hierarchical Negotiation Model for Complex Problems with Large-Number of Interdependent Issues

Negotiation is usually dealt with package deals, where all issues are considered together. We propose a hierarchical negotiation approach where various sets of issues are negotiated at different levels. As the number of interdependent issues in negotiation grows, it becomes computationally intractable to generate proposals on all issues and to evaluate such proposals. By modeling all issues in negotiation as a hierarchical structure, it significantly reduces the search space in each negotiation session and improves the probability of finding better negotiation outcomes. Since a numeric utility function for multiple interdependent issues is hard to acquire in most real-world applications, a constraint-based preference representation is adopted in this approach to mimic the preference model used by human. This paper presents the design of this hierarchical negotiation model, and illustrates how it works using an example of vehicle design process. Experimental results are presented to evaluate the the performance and efficiency of this approach.