Research On Strategy Of Argument-ation Negotiation Based On Adaptive Niche Genetic Algorithm

Nowdays, the traditional commercial negotiations could not meet the demand ofcommercial negotiations, so automatic negotiation based on Multi-Agent theory andtechnology would be widely applied. During the negotiation process, the buyer andseller do not only pursue the optimal individual interests, but also pay attention to theoverall interests.The researchers transform this particular question into mathematicalform by establishing negotiation models. In the process of solving the models, theyuse traditional machine learning algorithms, such as Genetic Algorithms, BayesianLearning, Particle Swarm Optimization Algorithm, etc. to process data. Theconvergence rates of traditional algorithms are rather slow, and final results tend to beprecocious too. The improved traditional machine learning algorithms could improvethe performance of the algorithm, which has become an important way to solve thisproblem. Using model solution only reflects the advantage on the outcome of thenegotiations, while ignoring the negotiations interaction, which is not effective inresolving the deadlock in the negotiations interaction. In this thesis, the theory ofargumentation is introduced to change the intentions and preferences of the negotiator,and break the impasse in the negotiations. Therefore, strategy of argumentation-basednegotiation has become the research focus.In view of these problems, the main topic of this paper is the selection strategy ofnegotiation and argumentation. The main contributions are as follows:Combine the current negotiations with multi-attribute utility theory as afoundation. In this thesis, adaptive niche genetic algorithm based negotiation model isdesigned, the model makes use of the Adaptive niche genetic algorithm on bilateralnegotiation to maximize the overall interests of negotiation solution for the globaloptimization, In this thesis, the best individual of each generation preservationstrategy is used. Dynamical distance function is designed, mutation rate and crossoverrate have been improved. The convergence rate of improved algorithm is faster, whichhas been verified by simulation experiments.In this thesis, the negotiation model based on multi-attribute is applied to obtainthe optimal solution. Argument preference level and case based reasoning method areintroduced to generate argument strategies and select argument strategies. By addingargumentation into negotiation, the agent’s belief and goal could be influenced dynamically, and negotiation proposal could be adjusted, which would make the finalproposal more optimal and the negotiation process more efficient and practical.Finally, according to the research, the argumentation-based multi-attributenegotiation system module is designed, which verifies that the model and strategiesare reasonable and practical.