A Heuristic Model for Concurrent Bi-Lateral Negotiations in Incomplete Information Settings

Abstract

<p>Multi-agent systems, in which autonomous agents interact in flexible ways, are an important new approach for developing software systems for a range of real-world problems. Here the notion of an agent (a computer program that is capable of autonomously working in its environment and interacting with other agents) is the core building block of the system. These software agents interact with one another in order to achieve their individual goals or to manage the dependencies that ensue from being situated in a common environment.</p> <p>This research developed a model that software agents can use to drive their participation in bilateral (pairwise) encounters. Specifically, we consider the case in which the agents negotiate over multiple issues (such as the price, quality and time of delivery) and where they can engage in multiple, concurrent encounters in order to procure the same good or service. The model is targeted at realistic trading scenarios (including web service procurement and virtual organization management) and so has to be computationally efficient and be able to operate effectively with minimal information about its negotiations opponents.</p> <p>To this end, we have developed a heuristic-based concurrent model that allows an agent to effectively handle simultaneous negotiations with other agents. A versatile coordination mechanism has been created to ensure that all the negotiations are inter-related to each other to ensure that only a single high value deal is reached at the end of the bargaining process. A commitment model has also been integrated to allow the agents to have more flexible behaviours and to stimulate different agents to participate in the negotiation process. Finally, an adaptive negotiation strategy has been introduced, which makes use of the information gained during the process, to improve the performance of the model in certain scenarios.</p>