|
Abstract:
|
Most of the time when someone wants to bargain over some good, service, or to negotiate over delicate matter, has a clear idea of what is wanted and not wanted as the negotiation outcome. There are deals that are totally unacceptable, some others that could be acceptable under some circumstances, and others that would be totally acceptable. Some times the most difficult part of a negotiation, is to ground the characteristics one wants as outcome. The system presented in this paper extracts the information that the user is expecting to get, by stating the characteristics, quantity and preference of each characteristic. Upon the preferences, a game is designed so a rational set of agents can solve the problem replacing the humans. A multi-agent system model that learns using learning classifier systems is shown to find negotiation solutions based on user preferences.
|
