Abstract

AbstractAutomated negotiation techniques play an important role in facilitating human in reaching better negotiation outcomes, and until now lots of research efforts have been devoted to designing effective negotiation strategies. To evaluate the performance of different strategies, one important evaluation criterion is robustness, which is to investigate which negotiating strategies the agents are going to adopt finally if they are given the opportunity to repeatedly negotiate and allowed to change their choices. However the current way of evaluating the robustness suffers from several drawbacks. First, it is assumed that all agents can have access to the global payoff information, which may not be available beforehand in practice. Second, it is based on the single-agent best deviation principle, however, in practice, each agent may change their strategies simultaneously and in any possible rational way. To this end, we firstly propose the repeated negotiation game learning framework to evaluate the robustness of different negotiation strategies, in which each agent can adopt any rational learning approach to make decisions without knowing the global payoff information beforehand. In this way, we are able to provide more realistic and fine-grained robustness analysis and more insights in terms of the relative robustness of different negotiating strategies can be revealed from our analytical results.KeywordsMixed StrategyPure StrategyRobustness AnalysisNegotiation StrategyStrategy Nash EquilibriumThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call