RandomDance: Compromising strategy considering interdependencies of issues with randomness

Abstract

In multi-lateral negotiations, agents need to simultaneously estimate the utility functions of more than two agents. In this chapter, we propose an estimating method that uses simple weighted functions by counting the opponent’s evaluation value for each issue. For multi-lateral negotiations, our agent considers some utility functions as the ‘single’ utility function by weighted-summing them. Our agent needs to judge which weighted function is effective and the types of the opponents. However, they depend on the domains, agents’ strategies and so on. Our agent selects the weighted function and opponent’s weighting, randomly.