Aprendizagem de máquina em um ambiente para negociações automatizadas.

Abstract

Automated negotiation is a hot research topic in AI applied to e-commerce. Particularly, one of the problems concerning the design of negotiation agents is to enclow them with learn ing and adaptation capabilities in order to face the dynamism often required by negotiation processes. This dissertation proposes an environment for building intelligent agents capable of ne gotiating and learning. Two experiments were done in order to demonstrate the feasibility of the proposed environment. The experiments show that the agents learned during the inter action process according to the used techniques: Neural networks, Q-learning, Rule-based Reasoning and Case-based Reasoning.