Evolving Optimal Agendas and Strategies for Negotiation in Dynamic Environments

Two key problems in a negotiation are: i) for the players to decide what issues to include in a negotiation, and ii) what strategy to use for negotiating over them. Let the number of available issues be m and let g be the number of issues to choose. The g issues thus chosen is called the negotiation agenda. An agent will choose the agenda that maximizes its utility and is therefore its optimal agenda. In many real-world negotiations, a player's actual utility from a deal will not be defined completely. Such scenarios make the problems of finding optimal agendas and strategies more challenging. In order to overcome this challenge, we present a multi surrogate-based GA system. This system is comprised of two GAs and set of Radial Basis Function Network (RBFN) surrogates that work together to find an optimal agenda and also an optimal strategy for that agenda.