This paper describes the agent-oriented simulation system to search for the mutual agreement point using an immune algorithm, for the purpose of the negotiation support. The utility theory is applied to explain the decision making process as a negotiation rule. The negotiation model is based on the concession with the agents correcting the utility function. The searching method of the mutual agreement point with the learning function of the immunity system was proposed, because the concession process expressed by the multi-step decision making could be replaced by the interaction between antibodies which compose the immunity network. As this method is modeled on the immunity network, it is able to search for the mutual agreement point of reliability. From the results of computer simulation, it has been understood to have the following advantages. (1) In the genetic algorithms, the complicated procedure of setting the initial value of the fitness was needed prior to solution searching. The method proposed here, however, can eliminate this procedure and it makes the algorithm much easier. (2) It can greatly decrease the number of candidates for solutions, and it can raise the processing efficiency in terms of the processing time and the storage area. (3) It can have the learning function corresponding for the neural network.
J-STAGEがリニューアルされました! https://www.jstage.jst.go.jp/browse/-char/ja/