A Multi-Issue Negotiation Mechanism for Bounded Rational Negotiators
Hamed Janzadeh, Ahmad Agha Kardan · 2008
So far, taking computational limitations into account just has been applied to single-issue negotiations while multiple-issue negotiations are more interesting because of further compatibility with the real world mechanisms. In this paper, we propose a mechanism for multiple-issue negotiations, where the agents have time limits to find the best contract. Every negotiation consists of several independent issues where finding the best-value for each of these issues becomes an intricate problem. The agents use an optimization algorithm to optimize their proposed contract. At every time step, computations are done by the agents to improve one of these negotiation issues. Since every negotiation has to be ended before a deadline, the negotiators might have no enough time to find the best contract. As a result, finding the best result requires precise management of time allocation in order to improve each issue of the negotiation. To solve this problem, our mechanism presents a deliberation control method to allocate a suitable computation time for each issue, thus causes the agents to find the best possible contract in allotted time. We consider two different types of the agent's knowledge about negotiation deadline. In the first type, agents are aware of each other's deadline, but in the second scenario every agent has a private deadline.