BDI Agent Architecture for Multi-Strategy Selection in Automated Negotiation

Abstract

Research in automated negotiation has traditionally been focusing on the negotiation protocol and strategy design, but little on the implementation related issues such as how to select the best negotiation strategy, especially for problems involving multi-strategy selection. The strategy selection is very important for agent to take risk to achieve better negotiation outcomes. The lack of such study has hampered the development in applying automated negotiation to real world problems. This research focuses on operationalizing risk taking agent's independent decision-making process through the design of a negotiation decision- making model and the software architecture, based on an abstract architecture model that can support both goal-directed reasoning and reactive response. We formally define the automated negotiation agent's abstract architecture model and propose an algorithm for the architecture and the decision-making model. Grounded on the theory of Belief-Desire-Intention, the model can support the agent's multi-strategy negotiation. A prototype of the model is built and applied to an aircraft purchase negotiation process to demonstrate the effectiveness of our model.