Bargaining strategies designed by evolutionary algorithms
Applied Soft Computing
| Hi-index | 0.00 |
Automated negotiation systems and real-world negotiation situationshave many aspects in common. Time is a relevant factor for allparties; information about preferences is private, and there is nointerest in having it disclosed; negotiators learn about theopponents and try to enhance their strategies while interactingwith one another. Experiments were performed with computationalagents employing a learning algorithm based on the ideas of theExperience-Weighted Attraction theory of learning in games, whichhas been shown to model well human behavior observed inexperimental settings. Negotiation strategies are acquired as theagents play bargaining games against one another. The strategiesdetermine the agents' behaviors: how much they offer to theopponent, when they make offers, and the conditions for acceptingan offer. The results show that the learning agents were able toacquire sensible strategies even from the most unstructured anddynamic environments. © 2006 Wiley Periodicals, Inc. Int J Int Syst 21: 539, 2006.