Aprendizagem de políticas de oferta de negociação entre agentes cognitivos
AUTOR(ES)
Emerson Romanhuki
DATA DE PUBLICAÇÃO
2008
RESUMO
This work proposes an approach to generate pro-active offer and counter-offers politics in a process of bilateral negotiations between cognitive agents with learning machine capacities. The applications of this work are destined to commercial negotiations to purchase and sale of products and services. The approach is configured, as each participant in a negotiation process can improve their satisfaction degree using knowledge obtained from previous experiences. Each participant individually refines their knowledge through the application of genetic algorithm on the historical dataset of offers with objective of redefine the parameters of configuration of the strategies and tactics. This approach was tested in a simulated bilateral negotiation environment, which involved a buyer agent and a seller agent. The discussions about the results confront different negotiations sessions, where the agents used and did not use the reconfiguration capabilities of their strategies and tactics to generation of offers and counter-offers. The values of the satisfaction were better when both participants of a negotiation session used the learning capacity and of adaptation. The learning of the best parameters group through genetic algorithm proved to be quite appropriate.
ASSUNTO(S)
algoritmos genéticos ciencia da computacao programação (computadores) informática - dissertações genetic programming (computer science) genetic algorithms programação genética (computação) computer programming
Documentos Relacionados
- Um modelo de negociação entre Bandwidth Brokers para oferta de QoS fim-a-fim
- Expert-Rete: um motor de inferência para agentes cognitivos
- Efeito de aprendizagem de testes cognitivos computadorizados em idosos
- Um modelo de negociação automatizada em comercio movel utilizando agentes moveis
- Negociação pedagógica aplicada a um ambiente multiagente de aprendizagem colaborativa